Ivan Frade [Thu, 28 Sep 2023 22:06:40 +0000 (15:06 -0700)]
UploadPack: Delay freeing refs in sendPack()
Change [1] set refs to null at the beggining of sendPack claiming they
are not needed anymore, but they are still used few lines below to
hoist referenced objects to the front of the pack. With refs nullified,
the hoist doesn't happened. This hasn't caused any problem so far,
probably because it is just an optimization and the objects are in the
pack anyway.
Move the nullification after the hoisting to keep the optimization and
save the memory.
Matthias Sohn [Sat, 23 Sep 2023 23:41:13 +0000 (01:41 +0200)]
WorkingTreeIterator: directly filter input stream
This way we can avoid to access the byte buffers backing array.
Implement a ByteBufferInputStream to wrap a byte buffer which we can use
to expose the filter result as an input stream.
See
https://docs.oracle.com/en/java/javase/11/docs/api/java.base/java/util/concurrent/CopyOnWriteArrayList.html
https://errorprone.info/bugpattern/ModifyCollectionInEnhancedForLoop
Matthias Sohn [Sat, 16 Sep 2023 20:19:41 +0000 (22:19 +0200)]
[errorprone] AddCommand#filepattern: use a more specific type
Variable type can use a more specific type to convey more information to
callers.
private Collection<String> filepatterns;
^
(see https://errorprone.info/bugpattern/PreferredInterfaceType)
Ivan Frade [Fri, 8 Sep 2023 20:23:09 +0000 (13:23 -0700)]
CommitGraphWriter: Decouple Stats from computing bloom filters
The public stats object is created only to be populated by the computation of
bloom filters.
Make the computation return its numbers with the results and copy them
to the stats when needed. This eliminates the side effects from the
computation and makes it easier to add more data to the stats later.
Ronald Bhuleskar [Mon, 18 Sep 2023 18:45:11 +0000 (11:45 -0700)]
Documentation: Move writeChangedPaths flag from commitGraph to gc section.
The flag commitGraph.writeChangedPaths is not correct. The option was introduced in commit 3b77e33 (Jul 19th) under gc section so updating the documentation to reflect it under correct section.
Matthias Sohn [Fri, 15 Sep 2023 09:48:05 +0000 (11:48 +0200)]
[errorprone] Fix wrong comparison which always evaluated to false
org.eclipse.jgit/src/org/eclipse/jgit/internal/storage/commitgraph/GraphObjectIndex.java:59:
error: [ComparisonOutOfRange] ints may have a value in the range
-2147483648 to 2147483647; therefore, this comparison to
Integer.MAX_VALUE will always evaluate to false
if (table[k] > Integer.MAX_VALUE) {
^
See https://errorprone.info/bugpattern/ComparisonOutOfRange
We need to check if variable `uint` of type `long` exceeds the maximum
possible int value before casting it to `int` below.
This was introduced in Ib5c0d6678cb242870a0f5841bd413ad3885e95f6
Matthias Sohn [Fri, 15 Sep 2023 09:44:09 +0000 (11:44 +0200)]
[errorprone] Remove unnecessary comparison
Raised by errorprone:
org.eclipse.jgit/src/org/eclipse/jgit/lib/CommitConfig.java:406: error:
[ComparisonOutOfRange] chars may have a value in the range 0 to 65535;
therefore, this comparison to 0 will always evaluate to true
if (ch >= 0 && ch < inUse.length) {
^
see https://errorprone.info/bugpattern/ComparisonOutOfRange
Matthias Sohn [Fri, 8 Sep 2023 20:57:05 +0000 (22:57 +0200)]
Use ShutdownHook to gracefully handle JVM shutdown
in all classes which already registered their own shutdown hook
- CloneCommand
- GC#PidLock
- FS#FileStoreAttributes
- LocalDiskRepositoryTestCase#Cleanup
Matthias Sohn [Fri, 8 Sep 2023 00:14:20 +0000 (02:14 +0200)]
Add ShutdownHook to cleanup FileLocks on graceful JVM shutdown
This should avoid stale lock files if the JVM is terminated gracefully.
Implement a ShutdownHook which can register/unregister listeners which
need to do some cleanup during graceful JVM shutdown. This hook is
registered as a Java shutdown hook and when the JVM shuts down
calls #onShutdown of registered listeners using a parallel stream
to let them run concurrently.
See https://docs.oracle.com/javase/8/docs/technotes/guides/lang/hook-design.html
Thomas Wolf [Mon, 11 Sep 2023 18:58:15 +0000 (20:58 +0200)]
OSGi: move plugin localization to subdirectory
OSGi can have its plugin localization at an arbitrary place; there is
no need to have it in a top-level plugin.properties file. In non-OSGi
environments having the files at the root level may mean that these
files clash with each other, or, as in the referenced bug, with some
third-party plug-in's plugin.properties, which may not even have
anything to do with localization.
Move our OSGi localization to a subfolder OSGI-INF/l10n. For OSGi
environments, that's just as good, and for non-OSGi environments it
avoid clashes with other root level items on the classpath or in a fat
JAR.
For fragments, use neither plugin.properties (which would clash with the
host plug-in's plugin.properties) nor fragment.properties (which might
clash with other fragments for the same fragment host bundle). Instead
use names "relative" to the host bundle.
Bug: 582394
Change-Id: Ifbcd046d912e2cfe86c0f7259c5ca8de599d9aa1 Signed-off-by: Thomas Wolf <twolf@apache.org>
Jonathan Nieder [Tue, 12 Sep 2023 00:12:17 +0000 (17:12 -0700)]
ssh: Remove redundant null check for home.getAbsoluteFile()
File#getAbsoluteFile is non-nullable, so this check can never trigger.
Worse, getAbsoluteFile can throw an exception such as
InvalidPathException, and since this call isn't in the "try" block
that checks for that, the exception would then escape the getSession
call.
Noticed because the exception is being thrown in googlesource.com's
custom SshdSessionFactory, causing incoming ssh requests to fail
(noticed using internal tests).
Change-Id: I57f2d5e497ff678b17573f79827b6e1d9a6c9b9f Signed-off-by: Jonathan Nieder <jrn@google.com>
Ivan Frade [Fri, 8 Sep 2023 15:35:47 +0000 (08:35 -0700)]
DfsPackFile: Record index loads only in one place
Each index can be set in the reader from two locations: the dfs cache
callback or the code afterwards. The pack is emitting the load event
in both cases, when the reference is set. This is brittle (right now
it is missing events for BITMAP_INDEX and COMMIT_GRAPH).
Emit the index loaded event only once, after going through the cache
code. The fact that the reference was set in the callback or the main
code is irrelevant. Also, the reader is per-thread, so there shouldn't
be any concurrency involved triggering double counts.
Ivan Frade [Fri, 8 Sep 2023 15:20:31 +0000 (08:20 -0700)]
DfsPackfile: Emit the index load with the index object
The DfsReader#emitIndexLoad expects a reference to the loaded object,
not to its identity hash. This makes listeneres recalculate an
identity hash over an Int object.
Matthias Sohn [Thu, 7 Sep 2023 00:23:57 +0000 (02:23 +0200)]
Merge branch 'stable-6.7'
* stable-6.7:
Prepare 6.7.1-SNAPSHOT builds
Switch back to wagon-ssh-external
Fix generation of maven site
JGit v6.7.0.202309050840-r
Remove unused API problem filters
[releng] Bump japicmp base version to 6.6.0.202305301015-r
Fix list of 3rd party bundles in p2 repo
Add missing source bundle org.osgi.service.cm.source to target platform
Prepare 6.6.2-SNAPSHOT builds
JGit v6.6.1.202309021850-r
Checkout: better directory handling
Document commit-graph options supported by JGit
Fix warning raised for local variable hiding DfsPackFile#index
Suppress boxing warnings in DfsPackFile
Remove unused API problem filters
Suppress boxing warnings in tests
Fix warning about empty block
Update mockito to 5.5.0
Update byte-buddy to 1.14.7
Prepare 6.7.0-SNAPSHOT builds
JGit v6.7.0.202308301100-rc1
Thomas Wolf [Fri, 11 Aug 2023 19:40:13 +0000 (21:40 +0200)]
Checkout: better directory handling
When checking out a file into the working tree ensure that all parent
directories of the file below the working tree root are actually
directories and do exist before we try to create the file.
When multiple files are to be checked out (or even a whole tree), this
may check the same directories over and over again. Asking the file
system every time for file attributes is a potentially expensive
operation. As a remedy, introduce an in-memory cache of directory
states for a particular check-out operation.
Apply the same fix also in the ResolveMerger, which may also check out
files, and also in the PatchApplier. In PatchApplier, also validate
paths.
Change-Id: Ie12864c54c9f901a2ccee7caddec73027f353111 Signed-off-by: Thomas Wolf <twolf@apache.org> Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Ivan Frade [Fri, 1 Sep 2023 18:25:50 +0000 (11:25 -0700)]
CommitGraphWriter: throw exception on unknown chunk
CommitGraphWriter first defines the chunks and then writes them. If at
write time a chunk is unknown, it is ignored. This is brittle: if
somebody adds a chunk to the header but not to the actual writing, the
commit-graph is broken and there is no error reported anywhere.
Throw exception if at write time a chunk is unknown. This can only
happen by a coding error in the writer.
projects / jgit.git / log