]> source.dussan.org Git - jgit.git/log
jgit.git
9 years agoMerge "Silence non-externalized string warnings in CherryPickResult"
Shawn Pearce [Wed, 8 Apr 2015 19:30:16 +0000 (15:30 -0400)]
Merge "Silence non-externalized string warnings in CherryPickResult"

9 years agoMerge "Use try-with-resource to close resources in CloneCommand"
Shawn Pearce [Wed, 8 Apr 2015 19:29:52 +0000 (15:29 -0400)]
Merge "Use try-with-resource to close resources in CloneCommand"

9 years agoMerge "Replace call to deprecated RevWalk.release() in DescribeCommand"
Shawn Pearce [Wed, 8 Apr 2015 19:26:19 +0000 (15:26 -0400)]
Merge "Replace call to deprecated RevWalk.release() in DescribeCommand"

9 years agoMerge "Use try-with-resource to close resources in DiffCommand"
Shawn Pearce [Wed, 8 Apr 2015 19:26:06 +0000 (15:26 -0400)]
Merge "Use try-with-resource to close resources in DiffCommand"

9 years agoMerge "Fix resource leak in DiffFormatter"
Shawn Pearce [Wed, 8 Apr 2015 19:25:29 +0000 (15:25 -0400)]
Merge "Fix resource leak in DiffFormatter"

9 years agoMerge "Use try-with-resource to close resources in DirCache"
Shawn Pearce [Wed, 8 Apr 2015 19:22:57 +0000 (15:22 -0400)]
Merge "Use try-with-resource to close resources in DirCache"

9 years agoMerge "Reduce visibility of DirCacheCheckout.checkValidPath()"
Shawn Pearce [Wed, 8 Apr 2015 19:21:15 +0000 (15:21 -0400)]
Merge "Reduce visibility of DirCacheCheckout.checkValidPath()"

9 years agoMerge "Use try-with-resource to close resources in FetchProcess"
Shawn Pearce [Wed, 8 Apr 2015 19:20:36 +0000 (15:20 -0400)]
Merge "Use try-with-resource to close resources in FetchProcess"

9 years agoMerge "Use try-with-resource to close resources in GC"
Shawn Pearce [Wed, 8 Apr 2015 19:20:13 +0000 (15:20 -0400)]
Merge "Use try-with-resource to close resources in GC"

9 years agoMerge "Replace deprecated release() methods by close() in Merger"
Shawn Pearce [Wed, 8 Apr 2015 19:19:18 +0000 (15:19 -0400)]
Merge "Replace deprecated release() methods by close() in Merger"

9 years agoMerge "Ensure that NameRevCommand.call() closes RevWalk"
Shawn Pearce [Wed, 8 Apr 2015 19:18:12 +0000 (15:18 -0400)]
Merge "Ensure that NameRevCommand.call() closes RevWalk"

9 years agoMerge "Replace deprecated release() methods by close() in NoteMapMerger"
Shawn Pearce [Wed, 8 Apr 2015 19:17:28 +0000 (15:17 -0400)]
Merge "Replace deprecated release() methods by close() in NoteMapMerger"

9 years agoMerge "Use try-with-resource to close resources in ObjectDatabase"
Shawn Pearce [Wed, 8 Apr 2015 19:17:11 +0000 (15:17 -0400)]
Merge "Use try-with-resource to close resources in ObjectDatabase"

9 years agoMerge changes Ia3908555,I3bc26847
Shawn Pearce [Wed, 8 Apr 2015 19:16:51 +0000 (15:16 -0400)]
Merge changes Ia3908555,I3bc26847

* changes:
  Silence false null pointer access warnings in PackFile
  Externalize error messages used in ObjectDirectoryInserter

9 years agoMerge "Silence non-externalized string warning in PackIndexV2"
Shawn Pearce [Wed, 8 Apr 2015 19:15:46 +0000 (15:15 -0400)]
Merge "Silence non-externalized string warning in PackIndexV2"

9 years agoMerge changes I1a4181fc,Ic2db0511
Shawn Pearce [Wed, 8 Apr 2015 19:15:32 +0000 (15:15 -0400)]
Merge changes I1a4181fc,Ic2db0511

* changes:
  Silence non-externalized string warning in PackParser
  Replace call to deprecated release() by close() in PackParser

9 years agoSilence non-externalized string warning in PackParser 84/45284/1
Matthias Sohn [Fri, 3 Apr 2015 22:54:24 +0000 (00:54 +0200)]
Silence non-externalized string warning in PackParser

Change-Id: I1a4181fc292c42157c69479f71c7674b2e00d2ce
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoReplace call to deprecated release() by close() in PackParser 83/45283/1
Matthias Sohn [Fri, 3 Apr 2015 22:53:47 +0000 (00:53 +0200)]
Replace call to deprecated release() by close() in PackParser

Change-Id: Ic2db051130059cce89486f55ecca32a7ce6715fd
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence non-externalized string warning in PackIndexV2 82/45282/1
Matthias Sohn [Fri, 3 Apr 2015 22:51:05 +0000 (00:51 +0200)]
Silence non-externalized string warning in PackIndexV2

Change-Id: I8c94811972c593c898f6375178f250a2123d01c6
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence false null pointer access warnings in PackFile 81/45281/1
Matthias Sohn [Fri, 3 Apr 2015 22:49:05 +0000 (00:49 +0200)]
Silence false null pointer access warnings in PackFile

Change-Id: Ia39085557b38840dfaa9b4995e6f6c40e19042cb
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoExternalize error messages used in ObjectDirectoryInserter 79/45279/1
Matthias Sohn [Fri, 3 Apr 2015 22:45:33 +0000 (00:45 +0200)]
Externalize error messages used in ObjectDirectoryInserter

Change-Id: I3bc26847071fbc31267a4a4cf5a10b428bcf229d
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in ObjectDatabase 77/45277/1
Matthias Sohn [Fri, 3 Apr 2015 22:38:39 +0000 (00:38 +0200)]
Use try-with-resource to close resources in ObjectDatabase

Change-Id: Ib410bf0d3c300c25b615bb6a51488b3d88aeb3bd
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoReplace deprecated release() methods by close() in NoteMapMerger 76/45276/1
Matthias Sohn [Fri, 3 Apr 2015 22:36:28 +0000 (00:36 +0200)]
Replace deprecated release() methods by close() in NoteMapMerger

Change-Id: Iaa76f0e853bb82cd7f231fd2f462a319f4196d8c
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoEnsure that NameRevCommand.call() closes RevWalk 75/45275/1
Matthias Sohn [Fri, 3 Apr 2015 22:32:59 +0000 (00:32 +0200)]
Ensure that NameRevCommand.call() closes RevWalk

Change-Id: Icc35a81081759d664fe9e2b05e962ff3cd9aad92
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoReplace deprecated release() methods by close() in Merger 72/45272/1
Matthias Sohn [Fri, 3 Apr 2015 22:19:39 +0000 (00:19 +0200)]
Replace deprecated release() methods by close() in Merger

Change-Id: I80ce98511a6e9048f1eea93288d6960dcbf3326f
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in GC 71/45271/1
Matthias Sohn [Fri, 3 Apr 2015 22:17:03 +0000 (00:17 +0200)]
Use try-with-resource to close resources in GC

Change-Id: I62a755a4ce839a252a5e80abf3f0d21243862376
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in FetchProcess 67/45267/1
Matthias Sohn [Fri, 3 Apr 2015 22:05:14 +0000 (00:05 +0200)]
Use try-with-resource to close resources in FetchProcess

Change-Id: If489d530ca39ae279c8da848f870b561c38eac3c
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoClearer error message when service is not enabled 26/43926/4
Hector Oswaldo Caballero [Mon, 16 Mar 2015 12:48:56 +0000 (08:48 -0400)]
Clearer error message when service is not enabled

When a user tried to use a service not enabled in the remote server
a misleading error  message was given:

   fatal: remote error: Git access forbidden

This patch modifies the error message to make the cause clearer
to the user. Now, when the user tries to use a not enabled service,
the message error clearly states it:

   fatal: remote error: Service not enabled

Change-Id: If096c4ddd17c5aae0e99e3ea6eea4b69bd3c5466
Signed-off-by: Hector Oswaldo Caballero <hector.caballero@ericsson.com>
9 years agoReduce visibility of DirCacheCheckout.checkValidPath() 98/45198/1
Matthias Sohn [Fri, 3 Apr 2015 00:03:00 +0000 (02:03 +0200)]
Reduce visibility of DirCacheCheckout.checkValidPath()

This was deprecated and should only be used by DirCacheCheckout and
friends. Other classes should use SystemReader.checkPath() instead.

Change-Id: I37cf753b1f081602dee9f0f47979eff39d735f92
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in DirCache 95/45195/1
Matthias Sohn [Thu, 2 Apr 2015 23:25:45 +0000 (01:25 +0200)]
Use try-with-resource to close resources in DirCache

Change-Id: I7dd3fd38784689816a1a8e9d8ccb499ae2138a8a
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix resource leak in DiffFormatter 94/45194/1
Matthias Sohn [Thu, 2 Apr 2015 23:23:31 +0000 (01:23 +0200)]
Fix resource leak in DiffFormatter

- close RevWalk allocated in scan()
- replace use of deprecated ObjectReader.release() method

Change-Id: I41b2b10a1a44270a6ceaa1741e996c0921439852
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in DiffCommand 93/45193/1
Matthias Sohn [Thu, 2 Apr 2015 23:15:37 +0000 (01:15 +0200)]
Use try-with-resource to close resources in DiffCommand

Change-Id: I13d7c8bd0eb049aa69effd8b74384746226414cb
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoReplace call to deprecated RevWalk.release() in DescribeCommand 90/45190/1
Matthias Sohn [Thu, 2 Apr 2015 22:58:50 +0000 (00:58 +0200)]
Replace call to deprecated RevWalk.release() in DescribeCommand

Change-Id: Ie4bfdeb37d345d6ff1525a737f3b7653b2f8e23e
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resource to close resources in CloneCommand 88/45188/1
Matthias Sohn [Thu, 2 Apr 2015 22:38:42 +0000 (00:38 +0200)]
Use try-with-resource to close resources in CloneCommand

Change-Id: I9536bc208a5f3ec34f0a82fb565b4253be38e074
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence non-externalized string warnings in CherryPickResult 86/45186/1
Matthias Sohn [Thu, 2 Apr 2015 22:36:36 +0000 (00:36 +0200)]
Silence non-externalized string warnings in CherryPickResult

Change-Id: I515abbcb4319b744c5074de5a0c56caed13c2fb5
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse try-with-resources to close walks in ArchiveCommand 88/45088/1
Matthias Sohn [Wed, 1 Apr 2015 22:58:13 +0000 (00:58 +0200)]
Use try-with-resources to close walks in ArchiveCommand

Change-Id: I77120d77a12f1bab5c918a23b0e3eac90e320b2b
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoBuffer overflow output stream 67/45067/1
Shawn Pearce [Wed, 1 Apr 2015 19:59:33 +0000 (12:59 -0700)]
Buffer overflow output stream

Most callers/users of TemporaryBuffer are sizing the in-memory
portion large enough that most outputs fit into RAM. With this
assumption they don't pay close attention to the size of IOs
being written, as it "should" just be a copy from one byte array
to another.

Overflow sets up a local file handle, which is costly to write to
for small IO units. Wrap the local file in a BufferedOutputStream
to combine small writes together. Larger writes can still bypass the
buffer as BOS automatically avoids copying for larger writes.

Change-Id: I09f4136dd65c48830cfda86d9101bc647581018a

9 years agoFix overflow stream leak during merge conflicts 66/45066/1
Shawn Pearce [Wed, 1 Apr 2015 19:57:06 +0000 (12:57 -0700)]
Fix overflow stream leak during merge conflicts

When reading back from an overflowed TemporaryBuffer the InputStream
must be closed to close the FileInputStream that is reading from
the backing file.

Change-Id: Id83d8f16f5b2c2618a9f841ec3508508455a6ae1

9 years agoUse local GIT_DIR for overflow during merge conflicts 65/45065/1
Shawn Pearce [Wed, 1 Apr 2015 19:53:00 +0000 (12:53 -0700)]
Use local GIT_DIR for overflow during merge conflicts

By writing the temporary overflow merge result to $GIT_DIR JGit
can ensure the same filesystem permissions apply to protect the
file contents.

If no directory is available from the repository (e.g. DfsRepository)
null will be passed and the system temporary directory will be used
instead.

Change-Id: I95532aa092676d18f1dc1e3fdbe6dcb1f91b782e

9 years agoUse try-with-resources and BufferedOutputStream for local merge conflicts 64/45064/1
Shawn Pearce [Wed, 1 Apr 2015 19:52:14 +0000 (12:52 -0700)]
Use try-with-resources and BufferedOutputStream for local merge conflicts

Change-Id: If5539aab1de19bb22400c862bbe2cdf7c3e85535

9 years agoOptimize EolAwareOutputStream for bulk output 63/45063/1
Shawn Pearce [Wed, 1 Apr 2015 19:47:57 +0000 (12:47 -0700)]
Optimize EolAwareOutputStream for bulk output

Formatting merge conflicts one byte at a time is going to be very
slow when the final OutputStream is a FileOutputStream and the JVM
is making system calls for each byte output.

When outputting a range of bytes from a byte[] the bol (beginning
of line) value only depends on the value of the last byte written.
Other bytes in the array can be passed directly to the lower stream
for more efficient output.

Change-Id: I3415f9a390ee215210a17bb5bf39164d197e1348

9 years agoResolveMerge only needs to visit differing TreeEntries 62/44962/1
Matthias Sohn [Tue, 31 Mar 2015 15:33:31 +0000 (17:33 +0200)]
ResolveMerge only needs to visit differing TreeEntries

This should considerably speed up the treewalk on larger repositories.

Found by discussing new EGit API to support model merge in change
eda23bb556d342f421f03b93c7faa160998598aa

Change-Id: I822721c76c64e614f87a080ced2457941f53adcd
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
cc: Laurent Delaigue <laurent.delaigue@obeo.fr>

9 years agoDo not add a newline at the end if neither merged side had one 19/33219/6
André de Oliveira [Wed, 10 Sep 2014 19:02:05 +0000 (16:02 -0300)]
Do not add a newline at the end if neither merged side had one

Bug: 390833
Change-Id: I29f7b79b241929877c93ac485c677487a91bb77b
Signed-off-by: André de Oliveira <andre.oliveira@liferay.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMake RepositoryState.REBASING_MERGE reachable again. 32/36832/4
Laurent Delaigue [Fri, 21 Nov 2014 10:28:19 +0000 (11:28 +0100)]
Make RepositoryState.REBASING_MERGE reachable again.

If a non interactive rebase is launched, stopping after a conflict
should set the repository state to RepositoryState.REBASING_MERGE
instead of RepositoryState.REBASING_INTERACTIVE.

Bug: 452623
Change-Id: Ie885aab6d71dabd158a718af0d14fff643c9b850
Also-by: Arthur Daussy <arthur.daussy@obeo.fr>
Signed-off-by: Laurent Delaigue <laurent.delaigue@obeo.fr>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix RecursiveMerger in case of multiple, independent base commits 67/44567/1
Christian Halstrick [Wed, 25 Mar 2015 09:48:03 +0000 (10:48 +0100)]
Fix RecursiveMerger in case of multiple, independent base commits

When RecursiveMerger found that there are multiple base-commits for the
commits to be merged it tries to temporarily merge the base commits. But
if these base commits have no common predecessor there was a bug in JGit
leading to a NPE. This commit fixes this by enforcing that an empty tree
is used as base when merging two unrelated base commits.

This logic was already there when merging two commits which have no
common predecessor (ThreeWayMerger.mergeBase()). But the code which was
computing a new temporary base commit in case of criss-cross merges
didn't take care to pick an empty tree when no common predecessor can be
found.

Bug: 462671
Change-Id: Ibd96302f5f81383f36d3b1e3edcbf5822147b1a4

9 years agoTestRepository: Add a cherryPick method 22/44122/4
Dave Borowitz [Wed, 18 Mar 2015 17:46:41 +0000 (10:46 -0700)]
TestRepository: Add a cherryPick method

CherryPickCommand only works on a non-bare repository, as it must
modify the working tree and index in case of a merge conflict. In
tests, being able to recover from a merge conflict is less important,
as the caller should be able to control the full contents of files in
advance of the cherry-pick.

Change-Id: Ic332e44df1308b9336e884666b08c1f6db64513d

9 years agoInclude slf4j and log4j in jgit command line 88/43488/3
Matthias Sohn [Mon, 9 Mar 2015 22:57:48 +0000 (15:57 -0700)]
Include slf4j and log4j in jgit command line

This enables the command line to log. Include log4j configuration to log
warnings and errors to stderr.

Exclude the dependencies which log4j 1.2.15 should have marked optional.
See
http://unitstep.net/blog/2009/05/18/resolving-log4j-1215-dependency-problems-in-maven-using-exclusions/
for details

Change-Id: Ie730db4007fb7614fd7d130cd0858b1ac550066a
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge bundle org.eclipse.jgit.java7 into org.eclipse.jgit 68/43768/2
Matthias Sohn [Fri, 13 Mar 2015 02:30:37 +0000 (19:30 -0700)]
Merge bundle org.eclipse.jgit.java7 into org.eclipse.jgit

As we moved minimum Java version to 7 we don't need a separate bundle
and feature for JGit features depending on Java 7 anymore.

Change-Id: Ib5da61b0886ddbdea65298f1e8c6d65c9879ced1
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd missing @since tag to DiffInterruptedException 67/44167/1
Matthias Sohn [Thu, 19 Mar 2015 13:53:43 +0000 (14:53 +0100)]
Add missing @since tag to DiffInterruptedException

Change-Id: Ibadbc1b476bfe0d1fe11979ec237a3554966eb59
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoRevert "CommitBuilder should check for duplicate parents" 36/44136/1
Jonathan Nieder [Wed, 18 Mar 2015 23:26:05 +0000 (16:26 -0700)]
Revert "CommitBuilder should check for duplicate parents"

This reverts commit 6bc48cdc62287934ce1b7003280b19a5994e7668.

Until git v1.7.10.2~29^2~1 (builtin/merge.c: reduce parents early,
2012-04-17), C git merge would make merge commits with duplicate parents
when asked to with a series of commands like the following:

  git checkout origin/master
  git merge --no-ff origin/master

Nowadays "git merge" removes redundant parents more aggressively
(whenever one parent is an ancestor of another and not just when
duplicates exist) but merges with duplicate parents are still permitted
and can be created with git fast-import or git commit-tree and history
viewers need to be able to cope with them.

CommitBuilder is an interface analagous to commit-tree, so it should
allow duplicate parents.  (That said, an option to automatically remove
redundant parents would be useful.)

Reported-by: Dave Borowitz <dborowitz@google.com>
Change-Id: Ia682238397eb1de8541802210fa875fdd50f62f0
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge changes I51167503,I794eca3a
Dave Borowitz [Wed, 18 Mar 2015 21:19:51 +0000 (17:19 -0400)]
Merge changes I51167503,I794eca3a

* changes:
  TemporaryBuffer: Clear block pointer list instead of reallocating
  TemporaryBuffer: Allow presizing block pointer list

9 years agoTemporaryBuffer: Clear block pointer list instead of reallocating 27/44127/3
Dave Borowitz [Wed, 18 Mar 2015 19:51:40 +0000 (12:51 -0700)]
TemporaryBuffer: Clear block pointer list instead of reallocating

The block pointer list may have been relatively large, so no need to
make more garbage. Instead, just clear the list and null out all the
elements.

Another possible motivation: a caller may have provided an inaccurate
estimated size, so the list might have been resized several times. If
the list is reused later for a similarly underestimated workload, this
fix will prevent additional resizing on subsequent usages.

Change-Id: I511675035dcff1117381a46c294cc11aded10893

9 years agoTemporaryBuffer: Allow presizing block pointer list 26/44126/3
Dave Borowitz [Wed, 18 Mar 2015 18:04:26 +0000 (11:04 -0700)]
TemporaryBuffer: Allow presizing block pointer list

Callers may wish to use TemporaryBuffer as an essentially unbounded
buffer by passing Integer.MAX_VALUE as the size. (This makes it
behave like ByteArrayOutputStream, only without requiring contiguous
memory.) Unfortunately, it was always allocating an array in the
backing block pointer list to hold enough blocks to MAX_VALUE--all
262,016 of them. It wasn't allocating the blocks themselves, but this
array was still extremely wasteful, using about 2MiB of memory on a
64-bit system.

Tweak the interface to specify an estimated size, and only allocate
the block pointer list enough entries to hold that size. It's an
ArrayList, so if that estimate was wrong, it'll grow. We assume the
cost of finding enough contiguous memory to grow that array is
acceptable.

While we're in there, fix an off-by-one error: due to integer division
we were undercounting the number of blocks needed to store n bytes of
data as (n / SZ).

Change-Id: I794eca3ac4472bcc605b3641e177922aca92b9c0

9 years agoMerge "Make MyersDiff interruptible"
Christian Halstrick [Wed, 18 Mar 2015 18:40:27 +0000 (14:40 -0400)]
Merge "Make MyersDiff interruptible"

9 years agoInMemoryRepository: Use a real Builder class 56/44056/2
Dave Borowitz [Wed, 18 Mar 2015 00:19:00 +0000 (17:19 -0700)]
InMemoryRepository: Use a real Builder class

Change-Id: I161b98a58503415955a21f2720395611f439ce98

9 years agoTestRepository: Expose a Git instance 51/44051/3
Dave Borowitz [Tue, 17 Mar 2015 22:15:45 +0000 (15:15 -0700)]
TestRepository: Expose a Git instance

Change-Id: I2fab7dd9a24205686db4724e77c0fd0a28f2075e

9 years agoGit: Don't close underlying repo if it came from from a caller 50/44050/2
Dave Borowitz [Tue, 17 Mar 2015 22:24:59 +0000 (15:24 -0700)]
Git: Don't close underlying repo if it came from from a caller

Since 27ae8bc65 Git has implemented AutoCloseable, which means Eclipse
may warn if close() is never called on a Git instance. For example,
the following would result in a resource warning:

  Repository repo = openRepository(foo);
  Git git = new Git(repo);
  try {
    git.someCommand().call();
  } finally {
    repo.close();
  }

(The same warning would occur if repo were created in a try-with-
resources block.)

The "obvious" fix is to open git in a try-with-resources block:

  try (Repository repo = openRepository(foo);
      Git git = new Git(repo)) {
    git.someCommand().call();
  }

Unfortunately, this construction was subtly broken: it would call both
git.close() and repo.close(), but git.close() would call repo.close()
again. Depending on the repository implementation, this might or might
not be ok. If it's not ok, it might not immediately cause an error, if
the reference count of repo was >2 at the time of closing.

Of course, explicitly calling git.close() followed by repo.close() in
two finally blocks has had the same double-closing problem since
forever. But the problem became worse when Git started implementing
AutoCloseable, because now Eclipse is _actively encouraging_
developers to change working code into broken code.

To work around this, keep track in Git's constructor of whether the
repository was passed in or opened at construction time, and only
close the repository if it was opened by Git.

Note that in the original example, there was not _actually_ a resource
leak, since repo was closed exactly once; git did not _need_ to be
closed in this case. But at least fixing this false-positive warning
no longer introduces a real bug.

Change-Id: Ie927a26ce3ae2bf8c3ef5cb963a60847067db95a

9 years agoMake MyersDiff interruptible 41/44041/1
Hugo Arès [Fri, 13 Mar 2015 17:57:38 +0000 (13:57 -0400)]
Make MyersDiff interruptible

For some specific file, MyersDiff goes into an infinite loop[1]. Since
this problem is hard to reproduce and possibly harder to fix, this
change makes the MyersDiff interruptible so the diff can be aborted at
least when such infinite loop happens.

[1]https://bugs.eclipse.org/bugs/show_bug.cgi?id=444623

Change-Id: I6e006ccb122d1e68c9846a24d5399d94776c2858
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
9 years agoFix string externalization warnings in BaseFormat 59/43759/2
Matthias Sohn [Thu, 12 Mar 2015 21:19:26 +0000 (14:19 -0700)]
Fix string externalization warnings in BaseFormat

Change-Id: Ie40aa1f889191e45e4d4a7a144c3176d521f6cfa
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd missing @since tags for new API methods in archive bundle 43/43743/2
Matthias Sohn [Thu, 12 Mar 2015 18:25:19 +0000 (11:25 -0700)]
Add missing @since tags for new API methods in archive bundle

Change-Id: I891e2cf9ca89ae1948e9713a412d31ec66faac86
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoCommitBuilder should check for duplicate parents 06/6606/5
Christian Halstrick [Wed, 4 Jul 2012 13:08:13 +0000 (15:08 +0200)]
CommitBuilder should check for duplicate parents

When setting the parents of a commit with setParentIds() or
addParentId() it should be checked that we don't have duplicate parents.
An IllegalArgumentException should be thrown in this case.

Change-Id: I9fa9f31149b7732071b304bca232f037146de454
Signed-off-by: Christian Halstrick <christian.halstrick@sap.com>
9 years agoCLI status should support --untracked-files 79/41279/4
Kaloyan Raev [Fri, 6 Feb 2015 14:45:58 +0000 (16:45 +0200)]
CLI status should support --untracked-files

A special options handler is added to properly handle the short -u alias
of the option.

The "normal" mode is not supported by this patch, because this mode of
listing untracked files is not
supported by the org.eclipse.jgit.lib.IndexDiff class. This mode is not
necessary for my use case. It can be added later if anyone really needs
it.

The StatusTest is updated to cover all possible combinations of the
--porcelain and --untracked-files options.

Bug: 459319
Change-Id: I305ac95739cfed0c16735e0987844e57fa27e236
Signed-off-by: Kaloyan Raev <kaloyan.r@zend.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoTestRepository: Allow setting explicit Change-Id 49/43749/1
Dave Borowitz [Thu, 12 Mar 2015 19:14:47 +0000 (12:14 -0700)]
TestRepository: Allow setting explicit Change-Id

This includes both leaving existing Change-Ids alone (as, for example
Gerrit's commit-msg hook does) and programmatically setting a value.

Change-Id: Iaaffb0107ae27de24df1f0e95a8d628fb8ea5364

9 years agoTestRepository: Ignore existing Change-Id in message 48/43748/1
Dave Borowitz [Thu, 12 Mar 2015 19:01:44 +0000 (12:01 -0700)]
TestRepository: Ignore existing Change-Id in message

Change-Id: I343c3065fa369ef8aab2b033ac1e392b28d10daf

9 years agoTestRepository: Support committing to unborn symref 42/43742/2
Dave Borowitz [Thu, 12 Mar 2015 17:59:02 +0000 (10:59 -0700)]
TestRepository: Support committing to unborn symref

Change-Id: I6bcee635adda7c929f6fa68ef20438fe77e49184

9 years agoTestRepository: Expose some getters for CommitBuilder 29/43729/5
Dave Borowitz [Thu, 12 Mar 2015 00:14:33 +0000 (17:14 -0700)]
TestRepository: Expose some getters for CommitBuilder

Change-Id: Ic6d179bd2de0081633c22fb82ca68ea619cb686f

9 years agoTestRepository: Add methods to amend commits or refs 00/43700/8
Dave Borowitz [Wed, 11 Mar 2015 19:04:10 +0000 (12:04 -0700)]
TestRepository: Add methods to amend commits or refs

Change-Id: I47082416f6e281262b160ba15272258f9109abd1

9 years agoTestRepository: Add a reset method to move HEAD around 01/43701/7
Dave Borowitz [Wed, 11 Mar 2015 22:21:48 +0000 (15:21 -0700)]
TestRepository: Add a reset method to move HEAD around

This flushed out a number of bugs in the way DfsRefUpdate, or at least
the InMemoryRepository implementation, processes symrefs. These have
been fixed, to an extent, in InMemoryRepository, but other
implementations may still suffer from these bugs.

Change-Id: Ifd12115a0060b9ff45a88d305b72f91ca0472f9a

9 years agoTestRepository: Optionally insert Change-Id in commit message 99/43699/3
Dave Borowitz [Wed, 11 Mar 2015 18:33:27 +0000 (11:33 -0700)]
TestRepository: Optionally insert Change-Id in commit message

Copied the implementation from CommitCommand.

Change-Id: Iade0e2d70bde70cfa830fe23bcc41959b011a14a

9 years agoMerge topic 'testrepo'
Shawn Pearce [Thu, 12 Mar 2015 17:11:20 +0000 (13:11 -0400)]
Merge topic 'testrepo'

* changes:
  TestRepository: Allow custom author/committer per-commit
  TestRepository: Use try-with-resources where appropriate

9 years agoFix compile error due to missing dependencies and since tags 14/43714/3
Markus Duft [Thu, 12 Mar 2015 10:04:02 +0000 (11:04 +0100)]
Fix compile error due to missing dependencies and since tags

Change-Id: I98a9f17f987c4f3ea19d107f681c44754ed83dca
Signed-off-by: Markus Duft <markus.duft@ssi-schaefer.com>
9 years agoTestRepository: Allow custom author/committer per-commit 98/43698/1
Dave Borowitz [Wed, 11 Mar 2015 18:28:15 +0000 (11:28 -0700)]
TestRepository: Allow custom author/committer per-commit

Change-Id: I078fe00470ebe60f93f4a718c163dd1412fdc776

9 years agoTestRepository: Use try-with-resources where appropriate 97/43697/1
Dave Borowitz [Wed, 11 Mar 2015 18:39:59 +0000 (11:39 -0700)]
TestRepository: Use try-with-resources where appropriate

Change-Id: I06f9534ab84278df37a140700fc2bed5ab667299

9 years agoMerge "ArchiveCommand: Allow to pass options to underlying stream"
Shawn Pearce [Wed, 11 Mar 2015 17:27:16 +0000 (13:27 -0400)]
Merge "ArchiveCommand: Allow to pass options to underlying stream"

9 years agoMerge changes I627681be,I334034a2
Shawn Pearce [Tue, 10 Mar 2015 23:55:18 +0000 (19:55 -0400)]
Merge changes I627681be,I334034a2

* changes:
  TreeWalk: Do not close reader passed explicitly to constructor
  TreeWalk: Stop using deprecated ObjectReader#release()

9 years agoTreeWalk: Do not close reader passed explicitly to constructor 17/43617/2
Dave Borowitz [Tue, 10 Mar 2015 22:21:06 +0000 (15:21 -0700)]
TreeWalk: Do not close reader passed explicitly to constructor

The TreeWalk(ObjectReader) constructor is explicitly to handle the case
where the caller is responsible for opening and closing the reader.
The reader should only be closed when it was created in the
TreeWalk(Repository) constructor.

Change-Id: I627681be80d69ea549f953255a64c7b3b68bcec9

9 years agoTreeWalk: Stop using deprecated ObjectReader#release() 16/43616/2
Dave Borowitz [Tue, 10 Mar 2015 22:24:47 +0000 (15:24 -0700)]
TreeWalk: Stop using deprecated ObjectReader#release()

Change-Id: I334034a2991a07664302bc8d1f3dead85c2caffe

9 years agoMake s3 domain dynamic to support different s3 regions 00/40600/2
Matthew Spurrier [Thu, 29 Jan 2015 10:19:58 +0000 (18:19 +0800)]
Make s3 domain dynamic to support different s3 regions

Change-Id: If8f9e85368c56d88bb6ae9efe1b3a29cc18cc1d5
Signed-off-by: Matthew Spurrier <matthew@spurrier.com.au>
9 years agoRevWalk: Do not close reader passed explicitly to constructor 12/43612/2
Dave Borowitz [Tue, 10 Mar 2015 22:21:06 +0000 (15:21 -0700)]
RevWalk: Do not close reader passed explicitly to constructor

The RevWalk(ObjectReader) constructor is explicitly to handle the case
where the caller is responsible for opening and closing the reader.
The reader should only be closed when it was created in the
RevWalk(Repository) constructor.

Change-Id: Ic0d595dc8d10de79e87549546c6c5ea2dc617e9b

9 years agoRevWalk: Stop using deprecated ObjectReader#release() 15/43615/1
Dave Borowitz [Tue, 10 Mar 2015 22:24:19 +0000 (15:24 -0700)]
RevWalk: Stop using deprecated ObjectReader#release()

Change-Id: If4d34f18352bd17467aeded6fd3478f29244657b

9 years agoUpdate all standard maven plugins to latest versions 10/43610/1
Dave Borowitz [Tue, 10 Mar 2015 21:33:43 +0000 (14:33 -0700)]
Update all standard maven plugins to latest versions

The latest versions of the javadoc and source plugins in particular
avoid some pathological slowness I'd been seeing on Linux with Java
7 or later.

Change-Id: I9fddd7e6ef513debec5f014ed2efc4fea6917d1f

9 years agoMerge "Add "--long" option to JGit describe"
Christian Halstrick [Tue, 10 Mar 2015 19:17:56 +0000 (15:17 -0400)]
Merge "Add "--long" option to JGit describe"

9 years agoRemove AutoCloseable from internal PackFile and friends 93/43493/2
Shawn Pearce [Tue, 10 Mar 2015 00:47:45 +0000 (17:47 -0700)]
Remove AutoCloseable from internal PackFile and friends

PackFile is held by the block cache and cannot be auto closed in a
try-with-resources statement.  Remove the interface as JGit does
explicit management of the instances.

ObjectDatabase and RefDatabase are internal details of Repository
and are managed with the Repository. Marking them AutoCloseable
provides no value to the library or an application using the API.

Change-Id: Ibee19eadd66233e6666b601583daa1834a7778f1

9 years agoPushCertificateParser: Fix check for blank line after header 92/43492/1
Shawn Pearce [Tue, 10 Mar 2015 00:30:08 +0000 (17:30 -0700)]
PushCertificateParser: Fix check for blank line after header

Reference equality (!= or ==) cannot be used to check for
String equality. String objects are not necessarily interned
to the same instance.

Use .isEmpty() since the function only cares about an empty
string and does not need to test a specific string value.

Change-Id: If530cb59666a8196d57d2348c893706a517ea541

9 years agoCleanup some push certificate related javadoc 91/43491/1
Shawn Pearce [Tue, 10 Mar 2015 00:28:16 +0000 (17:28 -0700)]
Cleanup some push certificate related javadoc

Change-Id: I319ee4e99462598bf6a934b1efc7939bc4b057a5

9 years agoFix an invalid format string 83/43483/3
David Pletcher [Mon, 9 Mar 2015 21:48:32 +0000 (14:48 -0700)]
Fix an invalid format string

The %x format specifier is not valid for a byte array.
This patch fixes a bug that would cause an IllegalFormatConversionException.

Change-Id: I025975eca7b2f10bbafa39f5519f8668e6536541
Signed-off-by: David Pletcher <dpletcher@google.com>
9 years agoAdd "--long" option to JGit describe 42/43142/5
Christian Halstrick [Wed, 4 Mar 2015 11:43:01 +0000 (12:43 +0100)]
Add "--long" option to JGit describe

Native git supports "git describe --long". This will enforce returning a
long description of a commit even if a tag is directly pointing to the
commit (in contrast to just returning the tag name as it is now). This
commit teaches JGits DescribeCommand and the describe command in the pgm
package to support "--long".

Bug: 460991
Change-Id: I65e179b79e89049c6deced3c71cb3ebb08ed0a8f
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Support for the commit-msg hook."
Matthias Sohn [Mon, 2 Mar 2015 23:17:54 +0000 (18:17 -0500)]
Merge "Support for the commit-msg hook."

9 years agoMerge "Allow public access to PackIndex"
Shawn Pearce [Mon, 2 Mar 2015 21:28:33 +0000 (16:28 -0500)]
Merge "Allow public access to PackIndex"

9 years agoAllow public access to PackIndex 36/42936/4
David Pletcher [Sat, 28 Feb 2015 01:17:52 +0000 (17:17 -0800)]
Allow public access to PackIndex

The index provides access to a list of objects in a pack.
This will be helpful for repository integrity checking.

Change-Id: I435eeeb3fe1b1f5632d40528936416e97491d412
Signed-off-by: David Pletcher <dpletcher@google.com>
9 years agoSupport for the commit-msg hook. 04/43004/2
Laurent Delaigue [Mon, 23 Feb 2015 10:19:45 +0000 (11:19 +0100)]
Support for the commit-msg hook.

This hook uses the file .git/COMMIT_EDITMSG to receive and potentially
modify the commit message.

Change-Id: Ibe2faadfb5d3932a5a3da2252d8156c4c04856c7
Signed-off-by: Laurent Delaigue <laurent.delaigue@obeo.fr>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoRefactored pre-commit hook to make it less invasive. 02/42402/8
Laurent Delaigue [Mon, 23 Feb 2015 10:18:50 +0000 (11:18 +0100)]
Refactored pre-commit hook to make it less invasive.

Hooks are now obtained via a convenient API like git commands, and
callers don't have to check for their existence.
The pre-commit hook has been updated accordingly.

Change-Id: I3383ffb10e2f3b588d7367b9139b606ec7f62758
Signed-off-by: Laurent Delaigue <laurent.delaigue@obeo.fr>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd an in-process pack transport for use in tests 34/42934/2
Dave Borowitz [Fri, 27 Feb 2015 23:08:50 +0000 (15:08 -0800)]
Add an in-process pack transport for use in tests

This allows for testing arbitrary sets of push/fetch hooks (e.g.
PreReceiveHook) without depending on either an external protocol (e.g.
HTTP) or the local filesystem.

Change-Id: I4ba2fff9c8a484f990dea05e14b0772deddb7411

9 years agoExtract classes for transport within a JGit process 33/42933/1
Dave Borowitz [Fri, 27 Feb 2015 23:04:35 +0000 (15:04 -0800)]
Extract classes for transport within a JGit process

TransportLocal knows how to spin up a thread to allow two repositories
in the same process to communicate using the wire protocol. However,
it is still tied to local on-disk filesystems, and needs to be able to
fork processes if not using the default git-{upload,receive}-pack
implementation.

Extract out the connection classes so they can be used by other
transport implementations.

Change-Id: I5db59086740735508c2e70a597c2d1a89014b072

9 years agoAdd missing since tags for new API 64/42864/1
Matthias Sohn [Fri, 27 Feb 2015 01:33:05 +0000 (02:33 +0100)]
Add missing since tags for new API

Change-Id: I90d4d7566b2e220b8ea8ea1c644fe4b7fa81c0a3
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'stable-3.7' 61/42861/1
Matthias Sohn [Fri, 27 Feb 2015 00:54:12 +0000 (01:54 +0100)]
Merge branch 'stable-3.7'

* stable-3.7:
  Prepare 3.7.1-SNAPSHOT builds
  JGit v3.7.0.201502260915-r
  Read user.name and email from environment first
  Provide more details in exceptions thrown when packfile is invalid

Change-Id: I427f861c6bc94da5e3e05dbbebbf0ad15719a323
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare 3.7.1-SNAPSHOT builds 60/42860/1
Matthias Sohn [Fri, 27 Feb 2015 00:41:30 +0000 (01:41 +0100)]
Prepare 3.7.1-SNAPSHOT builds

Change-Id: I2e97610ea9e552e5800e7ca895fd193c8bc507aa
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v3.7.0.201502260915-r 68/42768/1 v3.7.0.201502260915-r
Matthias Sohn [Thu, 26 Feb 2015 14:13:01 +0000 (15:13 +0100)]
JGit v3.7.0.201502260915-r

Change-Id: Iec17746cad81cfb1d775e782b30f9d8a13c938b6
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Read user.name and email from environment first" into stable-3.7
Christian Halstrick [Wed, 25 Feb 2015 17:01:40 +0000 (12:01 -0500)]
Merge "Read user.name and email from environment first" into stable-3.7

9 years agoAdd/fix since tags for new API for push certificates 32/42632/1
Matthias Sohn [Wed, 25 Feb 2015 13:18:12 +0000 (14:18 +0100)]
Add/fix since tags for new API for push certificates

This was missed in change I249869cadb2d55aef016371b9311b8583591b9cf

Change-Id: I19c9d4c04b6aa92b9e04c192dee70775d6985b58
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>