Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten Rewrite reference handling to be abstract and accurate
This commit actually does three major changes to the way references
are handled within JGit. Unfortunately they were easier to do as
a single massive commit than to break them up into smaller units.
Disambiguate symbolic references:
---------------------------------
Reporting a symbolic reference such as HEAD as though it were
any other normal reference like refs/heads/master causes subtle
programming errors. We have been bitten by this error on several
occasions, as have some downstream applications written by myself.
Instead of reporting HEAD as a reference whose name differs from
its "original name", report it as an actual SymbolicRef object
that the application can test the type and examine the target of.
With this change, Ref is now an abstract type with different
subclasses for the different types.
In the classical example of "HEAD" being a symbolic reference to
branch "refs/heads/master", the Repository.getAllRefs() method
will now return:
Map<String, Ref> all = repository.getAllRefs();
SymbolicRef HEAD = (SymbolicRef) all.get("HEAD");
ObjectIdRef master = (ObjectIdRef) all.get("refs/heads/master");
assertSame(master, HEAD.getTarget());
assertSame(master.getObjectId(), HEAD.getObjectId());
assertEquals("HEAD", HEAD.getName());
assertEquals("refs/heads/master", master.getName());
A nice side-effect of this change is the storage type of the
symbolic reference is no longer ambiguous with the storge type
of the underlying reference it targets. In the above example,
if master was only available in the packed-refs file, then the
following is also true:
assertSame(Ref.Storage.LOOSE, HEAD.getStorage());
assertSame(Ref.Storage.PACKED, master.getStorage());
(Prior to this change we returned the ambiguous storage of
LOOSE_PACKED for HEAD, which was confusing since it wasn't
actually true on disk).
Another nice side-effect of this change is all intermediate
symbolic references are preserved, and are therefore visible
to the application when they walk the target chain. We can
now correctly inspect chains of symbolic references.
As a result of this change the Ref.getOrigName() method has been
removed from the API. Applications should identify a symbolic
reference by testing for isSymbolic() and not by using an arcane
string comparsion between properties.
Abstract the RefDatabase storage:
---------------------------------
RefDatabase is now abstract, similar to ObjectDatabase, and a
new concrete implementation called RefDirectory is used for the
traditional on-disk storage layout. In the future we plan to
support additional implementations, such as a pure in-memory
RefDatabase for unit testing purposes.
Optimize RefDirectory:
----------------------
The implementation of the in-memory reference cache, reading, and
update routines has been completely rewritten. Much of the code
was heavily borrowed or cribbed from the prior implementation,
so copyright notices have been left intact as much as possible.
The RefDirectory cache no longer confuses symbolic references
with normal references. This permits the cache to resolve the
value of a symbolic reference as late as possible, ensuring it
is always current, without needing to maintain reverse pointers.
The cache is now 2 sorted RefLists, rather than 3 HashMaps.
Using sorted lists allows the implementation to reduce the
in-memory footprint when storing many refs. Using specialized
types for the elements allows the code to avoid additional map
lookups for auxiliary stat information.
To improve scan time during getRefs(), the lists are returned via
a copy-on-write contract. Most callers of getRefs() do not modify
the returned collections, so the copy-on-write semantics improves
access on repositories with a large number of packed references.
Iterator traversals of the returned Map<String,Ref> are performed
using a simple merge-join of the two cache lists, ensuring we can
perform the entire traversal in linear time as a function of the
number of references: O(PackedRefs + LooseRefs).
Scans of the loose reference space to update the cache run in
O(LooseRefs log LooseRefs) time, as the directory contents
are sorted before being merged against the in-memory cache.
Since the majority of stable references are kept packed, there
typically are only a handful of reference names to be sorted,
so the sorting cost should not be very high.
Locking is reduced during getRefs() by taking advantage of the
copy-on-write semantics of the improved cache data structure.
This permits concurrent readers to pull back references without
blocking each other. If there is contention updating the cache
during a scan, one or more updates are simply skipped and will
get picked up again in a future scan.
Writing to the $GIT_DIR/packed-refs during reference delete is
now fully atomic. The file is locked, reparsed fresh, and written
back out if a change is necessary. This avoids all race conditions
with concurrent external updates of the packed-refs file.
The RefLogWriter class has been fully folded into RefDirectory
and is therefore deleted. Maintaining the reference's log is
the responsiblity of the database implementation, and not all
implementations will use java.io for access.
Future work still remains to be done to abstract the ReflogReader
class away from local disk IO.
Change-Id: I26b9287c45a4b2d2be35ba2849daa316f5eec85d
Signed-off-by: Shawn O. Pearce <spearce@spearce.org>
14 vuotta sitten |
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380 |
- /*
- * Copyright (C) 2010, 2013 Google Inc.
- * and other copyright owners as documented in the project's IP log.
- *
- * This program and the accompanying materials are made available
- * under the terms of the Eclipse Distribution License v1.0 which
- * accompanies this distribution, is reproduced below, and is
- * available at http://www.eclipse.org/org/documents/edl-v10.php
- *
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or
- * without modification, are permitted provided that the following
- * conditions are met:
- *
- * - Redistributions of source code must retain the above copyright
- * notice, this list of conditions and the following disclaimer.
- *
- * - Redistributions in binary form must reproduce the above
- * copyright notice, this list of conditions and the following
- * disclaimer in the documentation and/or other materials provided
- * with the distribution.
- *
- * - Neither the name of the Eclipse Foundation, Inc. nor the
- * names of its contributors may be used to endorse or promote
- * products derived from this software without specific prior
- * written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
- * CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
- * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
- * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
- * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
- * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
- * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
- * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
- * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
- * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
- * ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- */
-
- package org.eclipse.jgit.lib;
-
- import java.io.IOException;
- import java.util.ArrayList;
- import java.util.Collection;
- import java.util.Collections;
- import java.util.HashMap;
- import java.util.List;
- import java.util.Map;
-
- /**
- * Abstraction of name to {@link ObjectId} mapping.
- * <p>
- * A reference database stores a mapping of reference names to {@link ObjectId}.
- * Every {@link Repository} has a single reference database, mapping names to
- * the tips of the object graph contained by the {@link ObjectDatabase}.
- */
- public abstract class RefDatabase {
- /**
- * Order of prefixes to search when using non-absolute references.
- * <p>
- * The implementation's {@link #getRef(String)} method must take this search
- * space into consideration when locating a reference by name. The first
- * entry in the path is always {@code ""}, ensuring that absolute references
- * are resolved without further mangling.
- */
- protected static final String[] SEARCH_PATH = { "", //$NON-NLS-1$
- Constants.R_REFS, //
- Constants.R_TAGS, //
- Constants.R_HEADS, //
- Constants.R_REMOTES //
- };
-
- /**
- * Maximum number of times a {@link SymbolicRef} can be traversed.
- * <p>
- * If the reference is nested deeper than this depth, the implementation
- * should either fail, or at least claim the reference does not exist.
- */
- protected static final int MAX_SYMBOLIC_REF_DEPTH = 5;
-
- /** Magic value for {@link #getRefs(String)} to return all references. */
- public static final String ALL = "";//$NON-NLS-1$
-
- /**
- * Initialize a new reference database at this location.
- *
- * @throws IOException
- * the database could not be created.
- */
- public abstract void create() throws IOException;
-
- /** Close any resources held by this database. */
- public abstract void close();
-
- /**
- * Determine if a proposed reference name overlaps with an existing one.
- * <p>
- * Reference names use '/' as a component separator, and may be stored in a
- * hierarchical storage such as a directory on the local filesystem.
- * <p>
- * If the reference "refs/heads/foo" exists then "refs/heads/foo/bar" must
- * not exist, as a reference cannot have a value and also be a container for
- * other references at the same time.
- * <p>
- * If the reference "refs/heads/foo/bar" exists than the reference
- * "refs/heads/foo" cannot exist, for the same reason.
- *
- * @param name
- * proposed name.
- * @return true if the name overlaps with an existing reference; false if
- * using this name right now would be safe.
- * @throws IOException
- * the database could not be read to check for conflicts.
- * @see #getConflictingNames(String)
- */
- public abstract boolean isNameConflicting(String name) throws IOException;
-
- /**
- * Determine if a proposed reference cannot coexist with existing ones. If
- * the passed name already exists, it's not considered a conflict.
- *
- * @param name
- * proposed name to check for conflicts against
- * @return a collection of full names of existing refs which would conflict
- * with the passed ref name; empty collection when there are no
- * conflicts
- * @throws IOException
- * @since 2.3
- * @see #isNameConflicting(String)
- */
- public Collection<String> getConflictingNames(String name)
- throws IOException {
- Map<String, Ref> allRefs = getRefs(ALL);
- // Cannot be nested within an existing reference.
- int lastSlash = name.lastIndexOf('/');
- while (0 < lastSlash) {
- String needle = name.substring(0, lastSlash);
- if (allRefs.containsKey(needle))
- return Collections.singletonList(needle);
- lastSlash = name.lastIndexOf('/', lastSlash - 1);
- }
-
- List<String> conflicting = new ArrayList<String>();
- // Cannot be the container of an existing reference.
- String prefix = name + '/';
- for (String existing : allRefs.keySet())
- if (existing.startsWith(prefix))
- conflicting.add(existing);
-
- return conflicting;
- }
-
- /**
- * Create a new update command to create, modify or delete a reference.
- *
- * @param name
- * the name of the reference.
- * @param detach
- * if {@code true} and {@code name} is currently a
- * {@link SymbolicRef}, the update will replace it with an
- * {@link ObjectIdRef}. Otherwise, the update will recursively
- * traverse {@link SymbolicRef}s and operate on the leaf
- * {@link ObjectIdRef}.
- * @return a new update for the requested name; never null.
- * @throws IOException
- * the reference space cannot be accessed.
- */
- public abstract RefUpdate newUpdate(String name, boolean detach)
- throws IOException;
-
- /**
- * Create a new update command to rename a reference.
- *
- * @param fromName
- * name of reference to rename from
- * @param toName
- * name of reference to rename to
- * @return an update command that knows how to rename a branch to another.
- * @throws IOException
- * the reference space cannot be accessed.
- */
- public abstract RefRename newRename(String fromName, String toName)
- throws IOException;
-
- /**
- * Create a new batch update to attempt on this database.
- * <p>
- * The default implementation performs a sequential update of each command.
- *
- * @return a new batch update object.
- */
- public BatchRefUpdate newBatchUpdate() {
- return new BatchRefUpdate(this);
- }
-
- /**
- * @return if the database performs {@code newBatchUpdate()} as an atomic
- * transaction.
- * @since 3.6
- */
- public boolean performsAtomicTransactions() {
- return false;
- }
-
- /**
- * Read a single reference.
- * <p>
- * Aside from taking advantage of {@link #SEARCH_PATH}, this method may be
- * able to more quickly resolve a single reference name than obtaining the
- * complete namespace by {@code getRefs(ALL).get(name)}.
- * <p>
- * To read a specific reference without using @{link #SEARCH_PATH}, see
- * {@link #exactRef(String)}.
- *
- * @param name
- * the name of the reference. May be a short name which must be
- * searched for using the standard {@link #SEARCH_PATH}.
- * @return the reference (if it exists); else {@code null}.
- * @throws IOException
- * the reference space cannot be accessed.
- */
- public abstract Ref getRef(String name) throws IOException;
-
- /**
- * Read a single reference.
- * <p>
- * Unlike {@link #getRef}, this method expects an unshortened reference
- * name and does not search using the standard {@link #SEARCH_PATH}.
- *
- * @param name
- * the unabbreviated name of the reference.
- * @return the reference (if it exists); else {@code null}.
- * @throws IOException
- * the reference space cannot be accessed.
- * @since 4.1
- */
- public Ref exactRef(String name) throws IOException {
- Ref ref = getRef(name);
- if (ref == null || !name.equals(ref.getName())) {
- return null;
- }
- return ref;
- }
-
- /**
- * Read the specified references.
- * <p>
- * This method expects a list of unshortened reference names and returns
- * a map from reference names to refs. Any named references that do not
- * exist will not be included in the returned map.
- *
- * @param refs
- * the unabbreviated names of references to look up.
- * @return modifiable map describing any refs that exist among the ref
- * ref names supplied. The map can be an unsorted map.
- * @throws IOException
- * the reference space cannot be accessed.
- * @since 4.1
- */
- public Map<String, Ref> exactRef(String... refs) throws IOException {
- Map<String, Ref> result = new HashMap<>(refs.length);
- for (String name : refs) {
- Ref ref = exactRef(name);
- if (ref != null) {
- result.put(name, ref);
- }
- }
- return result;
- }
-
- /**
- * Find the first named reference.
- * <p>
- * This method expects a list of unshortened reference names and returns
- * the first that exists.
- *
- * @param refs
- * the unabbreviated names of references to look up.
- * @return the first named reference that exists (if any); else {@code null}.
- * @throws IOException
- * the reference space cannot be accessed.
- * @since 4.1
- */
- public Ref firstExactRef(String... refs) throws IOException {
- for (String name : refs) {
- Ref ref = exactRef(name);
- if (ref != null) {
- return ref;
- }
- }
- return null;
- }
-
- /**
- * Get a section of the reference namespace.
- *
- * @param prefix
- * prefix to search the namespace with; must end with {@code /}.
- * If the empty string ({@link #ALL}), obtain a complete snapshot
- * of all references.
- * @return modifiable map that is a complete snapshot of the current
- * reference namespace, with {@code prefix} removed from the start
- * of each key. The map can be an unsorted map.
- * @throws IOException
- * the reference space cannot be accessed.
- */
- public abstract Map<String, Ref> getRefs(String prefix) throws IOException;
-
- /**
- * Get the additional reference-like entities from the repository.
- * <p>
- * The result list includes non-ref items such as MERGE_HEAD and
- * FETCH_RESULT cast to be refs. The names of these refs are not returned by
- * <code>getRefs(ALL)</code> but are accepted by {@link #getRef(String)}
- * and {@link #exactRef(String)}.
- *
- * @return a list of additional refs
- * @throws IOException
- * the reference space cannot be accessed.
- */
- public abstract List<Ref> getAdditionalRefs() throws IOException;
-
- /**
- * Peel a possibly unpeeled reference by traversing the annotated tags.
- * <p>
- * If the reference cannot be peeled (as it does not refer to an annotated
- * tag) the peeled id stays null, but {@link Ref#isPeeled()} will be true.
- * <p>
- * Implementors should check {@link Ref#isPeeled()} before performing any
- * additional work effort.
- *
- * @param ref
- * The reference to peel
- * @return {@code ref} if {@code ref.isPeeled()} is true; otherwise a new
- * Ref object representing the same data as Ref, but isPeeled() will
- * be true and getPeeledObjectId() will contain the peeled object
- * (or null).
- * @throws IOException
- * the reference space or object space cannot be accessed.
- */
- public abstract Ref peel(Ref ref) throws IOException;
-
- /**
- * Triggers a refresh of all internal data structures.
- * <p>
- * In case the RefDatabase implementation has internal caches this method
- * will trigger that all these caches are cleared.
- * <p>
- * Implementors should overwrite this method if they use any kind of caches.
- */
- public void refresh() {
- // nothing
- }
-
- /**
- * Try to find the specified name in the ref map using {@link #SEARCH_PATH}.
- *
- * @param map
- * map of refs to search within. Names should be fully qualified,
- * e.g. "refs/heads/master".
- * @param name
- * short name of ref to find, e.g. "master" to find
- * "refs/heads/master" in map.
- * @return The first ref matching the name, or null if not found.
- * @since 3.4
- */
- public static Ref findRef(Map<String, Ref> map, String name) {
- for (String prefix : SEARCH_PATH) {
- String fullname = prefix + name;
- Ref ref = map.get(fullname);
- if (ref != null)
- return ref;
- }
- return null;
- }
- }
|