]> source.dussan.org Git - jgit.git/log
jgit.git
9 years agoSilence deprecation warning in DirCacheCheckout 07/45707/7
Matthias Sohn [Sat, 11 Apr 2015 23:01:14 +0000 (01:01 +0200)]
Silence deprecation warning in DirCacheCheckout

Change-Id: I9e021ba2e0d1317bd98e86f832d55787ed6b0a63
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix hidden field warnings in bundle org.eclipse.jgit 06/45706/7
Matthias Sohn [Sat, 11 Apr 2015 22:51:37 +0000 (00:51 +0200)]
Fix hidden field warnings in bundle org.eclipse.jgit

Change-Id: I4e79cefe15037df0e1c520956bf2482240e31a7d
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix unchecked conversion warning in MergeFormatter 09/45709/7
Matthias Sohn [Sat, 11 Apr 2015 23:26:50 +0000 (01:26 +0200)]
Fix unchecked conversion warning in MergeFormatter

Change-Id: Id57ef8b2e80736652e4c5062364efa3ccd39c4c7
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoClose WindowCursor using try-with-resources in UnpackedObject 08/48908/1
Hugo Arès [Thu, 28 May 2015 19:59:48 +0000 (15:59 -0400)]
Close WindowCursor using try-with-resources in UnpackedObject

Change-Id: I187403ed432f7d6b1e84223918bd72d895e6bdfe
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
9 years agoMerge changes I7b6d7be4,I63a74651,I39c2ea6b
Shawn Pearce [Thu, 28 May 2015 19:39:07 +0000 (15:39 -0400)]
Merge changes I7b6d7be4,I63a74651,I39c2ea6b

* changes:
  Silence unused object warning in MyersDiff
  Silence resource leak warnings where caller is responsible to close
  Fix potential null pointer access in IndexDiffFilter

9 years agoMerge "Silence false potential null pointer access warnings"
Shawn Pearce [Thu, 28 May 2015 19:37:09 +0000 (15:37 -0400)]
Merge "Silence false potential null pointer access warnings"

9 years agoMerge "Fix WindowCursor memory leak."
Shawn Pearce [Thu, 28 May 2015 19:35:25 +0000 (15:35 -0400)]
Merge "Fix WindowCursor memory leak."

9 years agoMerge changes I144a53fe,I2b268e6a
Shawn Pearce [Thu, 28 May 2015 19:35:14 +0000 (15:35 -0400)]
Merge changes I144a53fe,I2b268e6a

* changes:
  Add tests for ObjectFilter
  Expose Sets helper to tests outside org.eclipse.jgit.api

9 years agoFix WindowCursor memory leak. 04/48904/1
Hugo Arès [Thu, 28 May 2015 18:21:03 +0000 (14:21 -0400)]
Fix WindowCursor memory leak.

ObjectReader release method was replaced by close method but
WindowCursor was still implementing release method.

To prevent the same mistake again, make ObjectReader close method
abstract to force sub classes to implement it.

Change-Id: I50d0d1d19a26e306fd0dba77b246a95a44fd6584
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
9 years agoarchive: Drop unnecessary empty comments and 'final' qualifiers on locals 02/48902/1
Jonathan Nieder [Wed, 27 May 2015 23:25:49 +0000 (16:25 -0700)]
archive: Drop unnecessary empty comments and 'final' qualifiers on locals

Early JGit code used comments to inform the Eclipse formatter about
where to break lines and used final in the hope of making code faster.
The ArchiveCommand command implementation imitated that style.

Nowadays the project relies less on the Eclipse formatter and relies
more on Java having sane performance with local variables that are not
explicitly marked 'final'.  Removing the unnecessary empty comments and
'final' qualifiers makes this code more readable and more consistent
with recent JGit code.

Change-Id: I7a181432eda7e18bd32cf110d89c0efbe490c4f1
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoClose 'out' consistently in ArchiveCommand.call 01/48901/1
Jonathan Nieder [Wed, 27 May 2015 23:25:32 +0000 (16:25 -0700)]
Close 'out' consistently in ArchiveCommand.call

Whether the output stream specified with setOutputStream() is closed by
ArchiveCommand.call() is murky and inconsistent:

- on success, it is closed
- if an exception is encountered when writing the archive, it is closed
- if an exception is encountered when calling createArchiveStream to
  open the archive, we forget to close it

Close the output stream consistently to avoid leaks.

Now that the inner try-with-resources doesn't have its own finally
block, this allows us to merge the two try blocks.

It would be even better to never close the output stream.  That will
involve more API changes to avoid silently breaking callers, so it is
deferred to a later change.

Change-Id: I0185bdaa60ecee4a541eab5d8ff6c9c4dbe40bf1
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoFix that exceptions in ReceivePack cause Invalid Channel 101 exceptions 12/48112/9
Christian Halstrick [Mon, 18 May 2015 22:45:42 +0000 (00:45 +0200)]
Fix that exceptions in ReceivePack cause Invalid Channel 101 exceptions

When during a PushOperation the server hits an exception different from
UnpackException the JGit server behaved wrong. That kind of exceptions
are handled so late that the connection is already released and the
information whether to talk sideband to the client is lost. In detail:
ReceivePack.receive() will call release() and that will reset the
capabilities. But later on the stack in ReceivePackServlet.doPost() it
is tried to send a response to client now with reset capabilities (no
sideband!).

Change-Id: I0a609acc6152ab43b47a93d712deb65bb1105f75
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoBetter report too large pack-files from PushCommand 48/46348/7
Christian Halstrick [Thu, 23 Apr 2015 13:24:25 +0000 (15:24 +0200)]
Better report too large pack-files from PushCommand

JGits PushCommand and BasePackPushConnection were throwing generic
TransportExceptions when the pushed pack-file was rejected by the server
since it was too big. Let JGit better interprete the server's response
to detect this situation and throw a more specific exception.

This detection works by parsing the status line sent by the server. This
change only recognizes the response sent by a JGit based server. All
other servers which report such problems in a different way still lead
to a generic TransportExceptions.

Change-Id: Ic075764ea152939ce72c446252464620dd54edea
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFS: Extract GobblerThread into a private static class 08/48808/2
Dave Borowitz [Wed, 27 May 2015 16:40:32 +0000 (09:40 -0700)]
FS: Extract GobblerThread into a private static class

The primary goal is to improve exception readability. Since this is a
standalone thread, just logging the stack trace of the caught
exception is not very useful:

java.io.IOException: Stream closed
at java.io.BufferedInputStream.getBufIfOpen(BufferedInputStream.java:162)
at java.io.BufferedInputStream.read(BufferedInputStream.java:258)
at org.eclipse.jgit.util.FS$2.run(FS.java:451)

Providing a named class eliminates the "FS$2", and including the
command name provides a little more context in the error message.
A future improvement might include the stack trace that created the
GobblerThread as well.

Change-Id: Ibf16d15b47a85b6f41844a177e398c2fc94f27b0

9 years agoAdd bitmap index misses to PackWriter.Statistics 92/48692/3
Terry Parker [Tue, 12 May 2015 00:37:02 +0000 (17:37 -0700)]
Add bitmap index misses to PackWriter.Statistics

RevWalks to find commits that are not in bitmap indices are expensive.
Track the count of commits that are enumerated via RevWalks as "bitmap
index misses" in the PackWriter.Statistics class.

Change-Id: Ie0135a0a0aeba2dfb6df78839d545006629f16cb
Signed-off-by: Terry Parker <tparker@google.com>
9 years agoMerge "Enable public access to SimilarityIndex scoring function"
Shawn Pearce [Tue, 26 May 2015 20:49:28 +0000 (16:49 -0400)]
Merge "Enable public access to SimilarityIndex scoring function"

9 years agoEnable public access to SimilarityIndex scoring function 74/48574/3
David Pletcher [Mon, 25 May 2015 21:49:58 +0000 (14:49 -0700)]
Enable public access to SimilarityIndex scoring function

The SimilarityIndex class implements the useful capability of scoring
the similarity between two files. That capability is required for a
feature that's being developed in another package, to detect files
derived from a set of potential sources.

This CL adds a public factory method to create a SimilarityIndex from
an ObjectLoader. It grants public access to the SimilarityIndex class,
the score method, an inner exception class and a special marker
instance of that exception class.

Change-Id: I3f72670da643be3bb8e261c5af5e9664bcd0401b
Signed-off-by: David Pletcher <dpletcher@google.com>
9 years agoMerge "Add getters to RepoProject."
Shawn Pearce [Tue, 26 May 2015 17:13:59 +0000 (13:13 -0400)]
Merge "Add getters to RepoProject."

9 years agoAdd getters to RepoProject. 55/48555/4
Yuxuan 'fishy' Wang [Mon, 25 May 2015 18:41:24 +0000 (11:41 -0700)]
Add getters to RepoProject.

Change-Id: I74ded6c2c3f5985568cd77bd8799b45017fb1d09
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoMerge branch 'stable-4.0' 53/48653/1
Matthias Sohn [Tue, 26 May 2015 14:54:18 +0000 (16:54 +0200)]
Merge branch 'stable-4.0'

* stable-4.0:
  Prepare post 4.0.0-rc2 builds
  JGit v4.0.0.201505260635-rc2

Change-Id: I1a67726b037be5311d85d90d815110b422182d33
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare post 4.0.0-rc2 builds 36/48636/1
Matthias Sohn [Tue, 26 May 2015 13:05:25 +0000 (15:05 +0200)]
Prepare post 4.0.0-rc2 builds

Change-Id: I3ba1bb0d7f220f88eb768a3137493f737aadf466
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v4.0.0.201505260635-rc2 12/48612/1 v4.0.0.201505260635-rc2
Matthias Sohn [Tue, 26 May 2015 10:33:24 +0000 (12:33 +0200)]
JGit v4.0.0.201505260635-rc2

Change-Id: I496743145da865f9631b46a432c65c1e63ccb501
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'master' into stable-4.0 11/48611/1
Matthias Sohn [Tue, 26 May 2015 09:39:19 +0000 (11:39 +0200)]
Merge branch 'master' into stable-4.0

* master:
  Silence non-externalized string warnings in org.eclipse.jgit
  Externalize translatable texts in org.eclipse.jgit
  Don't invalidate pack file on InterruptedIOException
  Update Mars target platforms to use Mars RC2 orbit
  Update build to use eclipse-jarsigner-plugin 1.1.2
  Guard agains null ReflogReader if named ref does not exist
  FS: Allow to manually set the path to the Git system config file
  FS: Fix a minor typo in runInShell() docs
  FS: Improve javadoc of some recently introduced methods
  Cleanup code and Eclipse compile errors in new gitrepo API
  Refactor to expose ManifestParser.
  FS: Remove the gitprefix logic
  SystemReader: Use discoverGitSystemConfig() in openSystemConfig()
  FS: Add a method to discover the system-wide config file
  FS: Extend readPipe() to optionally take additional environment
  FS: Document readpipe()'s encoding parameter
  Split discoverGitPrefix() code out into discoverGitExe()
  Equalize discoverGitPrefix() implementations between POSIX and Win32
  Move resolveGrandparentFile() to the base class for wider use
  Replace deprecated release() methods by close()
  Use AutoClosable to close resources in bundle org.eclipse.jgit
  ReceivePack: support quiet capability
  Fix ObjectReader resources leak

Change-Id: I0cd9f7ad57f26f0a0cbf412845d00ba1efbea346
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence unused object warning in MyersDiff 05/45705/5
Matthias Sohn [Sat, 11 Apr 2015 22:43:07 +0000 (00:43 +0200)]
Silence unused object warning in MyersDiff

Change-Id: I7b6d7be4bb283c1f451bc58f29c71a8082580989
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence resource leak warnings where caller is responsible to close 04/45704/5
Matthias Sohn [Fri, 10 Apr 2015 21:53:08 +0000 (23:53 +0200)]
Silence resource leak warnings where caller is responsible to close

Change-Id: I63a74651689c10426d5f150ab2e027c6b63cab95
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSilence false potential null pointer access warnings 00/45700/5
Matthias Sohn [Fri, 10 Apr 2015 00:37:04 +0000 (02:37 +0200)]
Silence false potential null pointer access warnings

Change-Id: I30c074a1b3436e47316b2eb2f083708161182d32

9 years agoSilence non-externalized string warnings in org.eclipse.jgit 05/45605/6
Matthias Sohn [Thu, 9 Apr 2015 22:56:45 +0000 (00:56 +0200)]
Silence non-externalized string warnings in org.eclipse.jgit

Change-Id: I7fa180db0c59e4a317a0c99ec48574cb0e5ac2f2
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix potential null pointer access in IndexDiffFilter 01/45701/5
Matthias Sohn [Sat, 11 Apr 2015 21:56:41 +0000 (23:56 +0200)]
Fix potential null pointer access in IndexDiffFilter

Change-Id: I39c2ea6b3090d7028cc1e246af6e3cc4e7bc70c2
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoExternalize translatable texts in org.eclipse.jgit 53/44053/8
Matthias Sohn [Tue, 17 Mar 2015 23:32:04 +0000 (00:32 +0100)]
Externalize translatable texts in org.eclipse.jgit

Change-Id: Ibf4c299f9d203c78cae79e61f88d4bea60ea2795
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoDon't invalidate pack file on InterruptedIOException 88/48288/3
Saša Živkov [Wed, 20 May 2015 14:57:24 +0000 (16:57 +0200)]
Don't invalidate pack file on InterruptedIOException

If the thread reading a pack file is interrupted don't invalidate that
pack file.

This could happen when Gerrit invoked JGit for computing a diff in one
thread and waited for the call to finish from another thread, with a
timeout. When the timeout was reached the "diff" thread was interrupted.
If it happened to be in an IO operation, reading a pack file, an
InterruptedIOException was thrown and the pack file was marked as
invalid and removed from the pack list.

Invalidating the pack in that case could cause the project disappearing in
Gerrit as discussed in [1] and [2].

[1] https://groups.google.com/forum/#!topic/repo-discuss/CYYoHfDxCfA
[2] https://groups.google.com/forum/#!topic/repo-discuss/ZeGWPyyJlrM

Change-Id: I2eb1f98370936b5be541d96d70c3973cbfc39238
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Signed-off-by: Sasa Zivkov <sasa.zivkov@sap.com>
9 years agoUpdate Mars target platforms to use Mars RC2 orbit 71/48571/1
Matthias Sohn [Mon, 25 May 2015 21:53:39 +0000 (23:53 +0200)]
Update Mars target platforms to use Mars RC2 orbit

Change-Id: Idda321d5026329ec194d53ad635973465bcccd49
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUpdate build to use eclipse-jarsigner-plugin 1.1.2 68/48568/1
Matthias Sohn [Mon, 25 May 2015 21:13:50 +0000 (23:13 +0200)]
Update build to use eclipse-jarsigner-plugin 1.1.2

Change-Id: Ic2fb30cf851542e12508e4cf638b52b65519f41d
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoGuard agains null ReflogReader if named ref does not exist 21/48521/3
Andrey Loskutov [Sun, 24 May 2015 07:15:15 +0000 (09:15 +0200)]
Guard agains null ReflogReader if named ref does not exist

Follow up on egit bug 466973.

Change-Id: Idd83d87803e86b25f106dfd725214b5a3ec5171c
Signed-off-by: Andrey Loskutov <loskutov@gmx.de>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFS: Allow to manually set the path to the Git system config file 77/48477/3
Sebastian Schuberth [Fri, 22 May 2015 15:21:27 +0000 (17:21 +0200)]
FS: Allow to manually set the path to the Git system config file

Now that d7a4473 removed the gitprefix property, we did not have a way to
specify the path to the Git system config file in case
discoverGitSystemConfig() fails. Fix that by introducing a member variable
that caches the result of discoverGitSystemConfig() as well as a setter
method to overwrite the content of that variable.

Change-Id: Icd965bffbe2f11b18c9505ee2ddd2afad5b64d70
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoFS: Fix a minor typo in runInShell() docs 57/48557/2
Sebastian Schuberth [Mon, 25 May 2015 18:07:53 +0000 (20:07 +0200)]
FS: Fix a minor typo in runInShell() docs

Change-Id: I5b1a5f3732aa97f83a4d39b42ba0f1e8eff9dec4
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoFS: Improve javadoc of some recently introduced methods 56/48556/2
Sebastian Schuberth [Mon, 25 May 2015 18:15:05 +0000 (20:15 +0200)]
FS: Improve javadoc of some recently introduced methods

Change-Id: I31e788ee20ac3e8439559d9060d39e9792f6dc7d
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoCleanup code and Eclipse compile errors in new gitrepo API 22/48522/3
Andrey Loskutov [Sun, 24 May 2015 07:31:17 +0000 (09:31 +0200)]
Cleanup code and Eclipse compile errors in new gitrepo API

Commit d3348e introduced few errors in Eclipse.

This commit cleans up the new API:
 - fixes API error in RepoCommand after moving IncludedFileReader type
 - fixes unused imports in RepoCommand & RepoCommandTest
 - fix javadoc errors in ManifestParser & RepoProject
 - makes three (implicitly final) fields in ManifestParser final.

Change-Id: I4185f451d97039d155391f62815bb9d3afe39fa6
Signed-off-by: Andrey Loskutov <loskutov@gmx.de>
9 years agoMerge "Refactor to expose ManifestParser."
Shawn Pearce [Fri, 22 May 2015 18:24:44 +0000 (14:24 -0400)]
Merge "Refactor to expose ManifestParser."

9 years agoRefactor to expose ManifestParser. 16/48416/2
Yuxuan 'fishy' Wang [Thu, 21 May 2015 23:52:32 +0000 (16:52 -0700)]
Refactor to expose ManifestParser.

The repo xml manifest parser used in RepoCommand could also be useful for
others, so refactor to make it public.

Also this breaks backward compatibility slightly.

Change-Id: I5001bd2fe77541109fe32dbe2597a065e6ad585e
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoFS: Remove the gitprefix logic 47/48147/6
Sebastian Schuberth [Tue, 19 May 2015 09:43:30 +0000 (11:43 +0200)]
FS: Remove the gitprefix logic

The only purpose of the gitprefix logic was to determine the path to the
system-wide config file. This is now done by discoverGitSystemConfig()
independent of the gitprefix, so get rid of this unused code.

Change-Id: Iaa88df9bd066dc1ed4067c18618af809e49876b3
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoSystemReader: Use discoverGitSystemConfig() in openSystemConfig() 46/48146/6
Sebastian Schuberth [Tue, 19 May 2015 09:30:03 +0000 (11:30 +0200)]
SystemReader: Use discoverGitSystemConfig() in openSystemConfig()

Bug: 410568
Change-Id: Id768294e83f374f50ae5e6486f4e757515c8262d
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFS: Add a method to discover the system-wide config file 45/48145/6
Sebastian Schuberth [Tue, 19 May 2015 08:49:44 +0000 (10:49 +0200)]
FS: Add a method to discover the system-wide config file

Change-Id: I969e26a5ab5f8ca3ab29024f405c1e34afdba493
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd tests for ObjectFilter 06/48006/4
Jonathan Nieder [Tue, 19 May 2015 01:28:31 +0000 (18:28 -0700)]
Add tests for ObjectFilter

Test that

 - the default ObjectFilter is ALL
 - ObjectFilter affects nextObject() and not next()
 - omitting a tree implies omitting its subtrees
 - a blob or tree reached by another path is still returned
 - ObjectFilter can be mixed with RevFilter

Change-Id: I144a53fe677070fff8c3ddf8cba07a848773bc1b
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoExpose Sets helper to tests outside org.eclipse.jgit.api 13/48413/1
Jonathan Nieder [Thu, 21 May 2015 23:15:06 +0000 (16:15 -0700)]
Expose Sets helper to tests outside org.eclipse.jgit.api

A later patch will make use of this class in a org.eclipse.jgit.lib
test.

Change-Id: I2b268e6a5dbf12174201f45259f9f007686708d2
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoFS: Extend readPipe() to optionally take additional environment 44/48144/5
Sebastian Schuberth [Tue, 19 May 2015 08:25:48 +0000 (10:25 +0200)]
FS: Extend readPipe() to optionally take additional environment

Change-Id: I4db7763826e4ada92074317d4d1c9a32299f3af8
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFS: Document readpipe()'s encoding parameter 43/48143/3
Sebastian Schuberth [Tue, 19 May 2015 08:21:59 +0000 (10:21 +0200)]
FS: Document readpipe()'s encoding parameter

Change-Id: I36f0edf0f9d0bfa26448e25355c6d78dea524079
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoSplit discoverGitPrefix() code out into discoverGitExe() 42/48142/3
Sebastian Schuberth [Tue, 19 May 2015 08:48:24 +0000 (10:48 +0200)]
Split discoverGitPrefix() code out into discoverGitExe()

Change-Id: I700540eec06efb24eeb09bfcb40420820c32d156
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoEqualize discoverGitPrefix() implementations between POSIX and Win32 41/48141/3
Sebastian Schuberth [Tue, 19 May 2015 07:35:18 +0000 (09:35 +0200)]
Equalize discoverGitPrefix() implementations between POSIX and Win32

Change-Id: I936df151890d4bba9079d79c65b75a69c209523b
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoMove resolveGrandparentFile() to the base class for wider use 40/48140/3
Sebastian Schuberth [Tue, 19 May 2015 07:30:02 +0000 (09:30 +0200)]
Move resolveGrandparentFile() to the base class for wider use

Change-Id: I67ec732ea2e5345a6946783f0c5ef60c07ce254e
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoReplace deprecated release() methods by close() 26/48326/4
Matthias Sohn [Fri, 10 Apr 2015 22:21:39 +0000 (00:21 +0200)]
Replace deprecated release() methods by close()

See the discussion [1] in the Gerrit mailing list.

[1] https://groups.google.com/forum/#!topic/repo-discuss/RRQT_xCqz4o

Change-Id: I2c67384309c5c2e8511a7d0d4e088b4e95f819ff
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse AutoClosable to close resources in bundle org.eclipse.jgit 03/45703/5
Matthias Sohn [Fri, 10 Apr 2015 22:24:27 +0000 (00:24 +0200)]
Use AutoClosable to close resources in bundle org.eclipse.jgit

- use try-with-resource where possible
- replace use of deprecated release() by close()

Change-Id: I0f139c3535679087b7fa09649166bca514750b81
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "ReceivePack: support quiet capability"
Matthias Sohn [Wed, 20 May 2015 21:49:12 +0000 (17:49 -0400)]
Merge "ReceivePack: support quiet capability"

9 years agoReceivePack: support quiet capability 41/48241/2
Shawn Pearce [Wed, 20 May 2015 07:34:00 +0000 (00:34 -0700)]
ReceivePack: support quiet capability

git-core has supported this for a long time; allowing clients to
avoid progress messages from the server if they are dumping to a
pipe instead of a tty.

Avoid the two progress monitors going on side-band and expose
isQuiet() method to allow hooks to also reduce their output if
this is sensible for them.

Change-Id: I1df7e38d16765446b441366500b017a90b8ff958

9 years agoMerge branch 'stable-4.0' 39/48239/1
Matthias Sohn [Wed, 20 May 2015 06:55:38 +0000 (08:55 +0200)]
Merge branch 'stable-4.0'

* stable-4.0:
  Prepare post 4.0.0-rc1 builds
  JGit v4.0.0.201505191015-rc1

Change-Id: I9731a591368e2e7ed642e36bd1ffee312b9568c7
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Fix ObjectReader resources leak"
Shawn Pearce [Wed, 20 May 2015 06:02:25 +0000 (02:02 -0400)]
Merge "Fix ObjectReader resources leak"

9 years agoPrepare post 4.0.0-rc1 builds 20/48220/1
Matthias Sohn [Tue, 19 May 2015 20:34:37 +0000 (22:34 +0200)]
Prepare post 4.0.0-rc1 builds

Change-Id: I5c18bb97336e5b94fae46743cbbb0f02a6e4b477
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v4.0.0.201505191015-rc1 76/48176/1 v4.0.0.201505191015-rc1
Matthias Sohn [Tue, 19 May 2015 14:12:20 +0000 (16:12 +0200)]
JGit v4.0.0.201505191015-rc1

Change-Id: Ib3371a082cbfaafd67471c56807b6d89f1161c8b
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'master' into stable-4.0 75/48175/1
Matthias Sohn [Tue, 19 May 2015 14:07:17 +0000 (16:07 +0200)]
Merge branch 'master' into stable-4.0

* master:
  Fix warnings in ObjectFilter
  Fix typo in reflog message written by RebaseCommand.tryFastForward()
  Correct @since tags for ObjectFilter
  Fix typo in ObjectWalk#getObjectFilter javadoc
  Allow ObjectWalk to be filtered by an arbitrary predicate
  Remove SoftReference from dfs.DeltaBaseCache
  Fix memory leak in dfs.DeltaBaseCase
  Update to Jetty 9.2.10
  Update javax.servlet to 3.1
  Use ANY_DIFF filter in ResolveMerger only for bare repositories
  FS_POSIX: Rework umask detection to make it settable
  Expose disposeBody() on RevCommit and RevTag
  ObjectReader: remove the walkAdvice API
  RevWalk: Discard uninteresting commits unless RevSort.BOUNDARY
  ObjectWalk: make setRetainBody(false) the default
  Do not concatenate strings as arguments to StringBuilder.append()
  IndexDiffFilter: Simplify a boolean expression
  GroupHead: Remove a redundant call to String.format()
  FS_Win32: Avoid an IOException on Windows if bash is not in PATH
  Skip logging stack trace on corrupt objects
  Add repository name to failures in HTTP server log
  Fix possible AIOOB in DirCacheTree.contains()
  Delete deprecated PackWriter.preparePack() methods
  Delete deprecated class IgnoreRule
  Delete deprecated checkoutEntry() methods in DirCacheCheckout
  Fix IllegalArgumentException in AmazonS3

Change-Id: Ica3d4f0675c81684fbe48fcf0053f2a949bc5c9b
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix warnings in ObjectFilter 98/47998/1
Matthias Sohn [Fri, 15 May 2015 21:20:46 +0000 (23:20 +0200)]
Fix warnings in ObjectFilter

- add missing tags in JavaDoc of ObjectFilter.include()
- remove unnecessary import

Change-Id: I24b9dcc49f66380f77345d704df70c05f7f74db8
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix typo in reflog message written by RebaseCommand.tryFastForward() 66/47966/2
Matthias Sohn [Fri, 15 May 2015 13:54:38 +0000 (15:54 +0200)]
Fix typo in reflog message written by RebaseCommand.tryFastForward()

Change-Id: I1ad544f2b5673ed3b4a2206b5eb4ce20fd3c86d2
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFix ObjectReader resources leak 93/47993/1
Hugo Arès [Fri, 15 May 2015 19:15:28 +0000 (15:15 -0400)]
Fix ObjectReader resources leak

In 77030a5e, AutoClosable was implemented on classes that use release().
This caused a resource leak because the ObjectReader.close method  was
not calling the now deprecated release method, which is the method that
sub classes implements to release resources.

Change-Id: I247651ec8fd7ca9941d256ca46d14cc43cc35c6e
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
9 years agoCorrect @since tags for ObjectFilter 90/47990/1
Jonathan Nieder [Fri, 15 May 2015 18:37:57 +0000 (11:37 -0700)]
Correct @since tags for ObjectFilter

Although the stable-4.0 branch already exists, 4.0 development is
still happening on master until IP logs are sent for review, which
will happen at the end of May.

Change-Id: I863ba85c6303f8ef2eb13bca5e2d30e5d3c58b29
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoFix typo in ObjectWalk#getObjectFilter javadoc 89/47989/1
Jonathan Nieder [Fri, 15 May 2015 18:23:55 +0000 (11:23 -0700)]
Fix typo in ObjectWalk#getObjectFilter javadoc

While trying to decide between "which matches every object" and "as it
matches every object", I became distracted and wrote both.

Change-Id: I867ce29664e661a81a9d441e59ffd0b72270dd98
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoAllow ObjectWalk to be filtered by an arbitrary predicate 38/47938/4
Jonathan Nieder [Thu, 14 May 2015 23:24:15 +0000 (16:24 -0700)]
Allow ObjectWalk to be filtered by an arbitrary predicate

This will make it possible to declare a collection of objects as
ineligible for the walk en masse, for example if they are known to be
uninteresting via a bitmap.

Change-Id: I637008b25bf9fb57df60ebb2133a70214930546a
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge "Update to Jetty 9.2.10"
Matthias Sohn [Tue, 12 May 2015 13:00:01 +0000 (09:00 -0400)]
Merge "Update to Jetty 9.2.10"

9 years agoRemove SoftReference from dfs.DeltaBaseCache 70/47570/2
Shawn Pearce [Sun, 10 May 2015 02:53:53 +0000 (19:53 -0700)]
Remove SoftReference from dfs.DeltaBaseCache

The Java GC doesn't always clear these before running out of memory
and failing allocations. In practice OpenJDK 7 is leaving these live,
removing any advantage of the SoftReference to attempt to shed memory
when the GC is unable to continue allocating.

Instead follow the pattern of the DfsBlockCache and use hard refs
to the object data. Require applications to configure the cache
size more accurately given expected memory usage.

Change-Id: I87586b3e71b1cba0308a6a278d42e971be4bccd3

9 years agoFix memory leak in dfs.DeltaBaseCase 69/47569/2
Shawn Pearce [Sun, 10 May 2015 01:11:20 +0000 (18:11 -0700)]
Fix memory leak in dfs.DeltaBaseCase

The LRU chain management code was broken leading to situations where
the chain was incomplete.  This prevented the cache from removing
items when it exceeded its memory target, causing a leak.

One case was repeated hit on the head of the chain. moveToHead(e)
was invoked linking the head back to itself in a cycle orphaning
the rest of the table.

Add some unit tests to cover this and a few other paths.

Change-Id: Ib27486eaa1b1d2bf1c745a56d0a5832bfb029322

9 years agoMerge "FS_POSIX: Rework umask detection to make it settable"
Shawn Pearce [Mon, 11 May 2015 18:17:46 +0000 (14:17 -0400)]
Merge "FS_POSIX: Rework umask detection to make it settable"

9 years agoMerge "Fix possible AIOOB in DirCacheTree.contains()"
Christian Halstrick [Mon, 11 May 2015 08:09:53 +0000 (04:09 -0400)]
Merge "Fix possible AIOOB in DirCacheTree.contains()"

9 years agoUpdate to Jetty 9.2.10 73/47573/6
Matthias Sohn [Sun, 10 May 2015 07:48:38 +0000 (09:48 +0200)]
Update to Jetty 9.2.10

Change-Id: Iace29e6e99836019bb603ce06a08b91bada7c627
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUpdate javax.servlet to 3.1 72/47572/4
Matthias Sohn [Fri, 8 May 2015 22:56:43 +0000 (00:56 +0200)]
Update javax.servlet to 3.1

Change-Id: Ifad154ed2f52f0102d297ac7fd4943ff1f309b9e
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUse ANY_DIFF filter in ResolveMerger only for bare repositories 18/45718/4
Matthias Sohn [Sun, 12 Apr 2015 23:01:13 +0000 (01:01 +0200)]
Use ANY_DIFF filter in ResolveMerger only for bare repositories

As Chris pointed out change I822721c76c64e614f87a080ced2457941f53adcd
slowed down merge since ANY_DIFF filter is much less efficient than the
manual detection of diffs done in ResolveMerger.processEntry() since it
avoids unnecessary filesystem calls using the git index. Hence only set
the ANY_DIFF filter on bare repositories which don't have a working tree
to scan.

To test performance I used the setup described in Chris' comment on
change I822721c76c64e614f87a080ced2457941f53adcd and modified
ResolveMerger.mergeTrees() to not add the working tree in order to
simulate merging in a bare repository.

At least on Mac I couldn't detect a speedup, with and without the
ANY_DIFF filter merge test takes an average 0.67sec.

Change-Id: I17b3a06f369cee009490f54ad1a2deb6c145c7cf
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoFS_POSIX: Rework umask detection to make it settable 87/47587/3
Shawn Pearce [Sun, 10 May 2015 19:19:12 +0000 (12:19 -0700)]
FS_POSIX: Rework umask detection to make it settable

Avoid always calling `sh -c umask` on startup, instead deferring
the invocation until the first time a working tree file needs to
use the execute bit. This allows servers using bare repos to avoid
a costly fork+exec for a value that is never used.

Store the umask as an int instead of two Boolean. This is slightly
smaller memory (one int vs. two references) and makes it easier for
an application to force setting the umask to a value that overrides
whatever the shell told JGit.

Simplify the code to bail by returning early when canExecute is
false, which is the common case for working tree files.

Change-Id: Ie713647615bc5bdf5d71b731a6748c28ea21c900

9 years agoMerge "Fix IllegalArgumentException in AmazonS3"
Shawn Pearce [Sun, 10 May 2015 18:44:15 +0000 (14:44 -0400)]
Merge "Fix IllegalArgumentException in AmazonS3"

9 years agoExpose disposeBody() on RevCommit and RevTag 85/47585/2
Shawn Pearce [Sun, 10 May 2015 17:42:09 +0000 (10:42 -0700)]
Expose disposeBody() on RevCommit and RevTag

Applications that use a commit message once and do not
need it again can free the body to save memory. Expose
the disposeBody() methods to support this and use it in
pgm.Log which only visits each commit once.

Change-Id: I4142a0749c24f15386ee7fb119934a0432234de3

9 years agoObjectReader: remove the walkAdvice API 61/47561/5
Shawn Pearce [Sat, 9 May 2015 05:26:28 +0000 (22:26 -0700)]
ObjectReader: remove the walkAdvice API

This was added a very long time ago to support the failed
DHT storage implementation. Since then no storage system
was able to make use of this API, but it pollutes internals
of the walkers.

Kill the API on ObjectReader and drop the invocations from
the walker code.

Change-Id: I36608afdac13a6c3084d7c7e0af5e0cb22900332

9 years agoRevWalk: Discard uninteresting commits unless RevSort.BOUNDARY 63/47563/4
Shawn Pearce [Sat, 9 May 2015 05:48:10 +0000 (22:48 -0700)]
RevWalk: Discard uninteresting commits unless RevSort.BOUNDARY

Previously using an ObjectWalk meant uninteresting commits may keep
their commit message buffers in memory just in case they were found to
be on the boundary and were output as UNINTERESTING for the caller.

This was incorrect inside StartGenerator. ObjectWalk hides these
internal UNINTERESTING cases from its caller unless RevSort.BOUNDARY
was explicitly set, and its false by default. Callers never see one
of these saved uninteresting commits.

Change the test to allow early dispose unless the application has
explicitly asked for RevSort.BOUNDARY. This allows uninteresting
commit buffers to be discarded and garbage collected in ObjectWalks
when the caller will never be given the RevCommit.

Change-Id: Ic1419cc1d9ee95f4d09386dd0730d54c12dcc157

9 years agoObjectWalk: make setRetainBody(false) the default 68/47568/3
Shawn Pearce [Sat, 9 May 2015 17:47:13 +0000 (10:47 -0700)]
ObjectWalk: make setRetainBody(false) the default

Despite being the primary author of RevWalk and ObjectWalk I still
fail to remember to setRetainBody(false) in application code using
an ObjectWalk to examine the graph.

Document the default for RevWalk is setRetainBody(true), where the
application usually wants the commit bodies to display or inspect.

Change the default for ObjectWalk to setRetainBody(false), as nearly
all callers want only the graph shape and do not need the larger text
inside a commit body. This allows some code in JGit to be simplified.

Change-Id: I367e42209e805bd5e1f41b4072aeb2fa98ec9d99

9 years agoMerge changes I62befa4a,Ia9c3d123,I7fcf6534
Shawn Pearce [Sun, 10 May 2015 03:00:55 +0000 (23:00 -0400)]
Merge changes I62befa4a,Ia9c3d123,I7fcf6534

* changes:
  Delete deprecated PackWriter.preparePack() methods
  Delete deprecated class IgnoreRule
  Delete deprecated checkoutEntry() methods in DirCacheCheckout

9 years agoDo not concatenate strings as arguments to StringBuilder.append() 23/47523/1
Sebastian Schuberth [Fri, 8 May 2015 09:04:27 +0000 (11:04 +0200)]
Do not concatenate strings as arguments to StringBuilder.append()

That more or less defeats the purpose of using a StringBuilder.

Change-Id: I519f7bf1c9b6670e63c3714210f834ee845dc69f
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoIndexDiffFilter: Simplify a boolean expression 22/47522/1
Sebastian Schuberth [Fri, 8 May 2015 09:02:44 +0000 (11:02 +0200)]
IndexDiffFilter: Simplify a boolean expression

Change-Id: Ibdd0338b638b864d6572045b084b08a04471ecf7
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoGroupHead: Remove a redundant call to String.format() 21/47521/1
Sebastian Schuberth [Fri, 8 May 2015 09:02:15 +0000 (11:02 +0200)]
GroupHead: Remove a redundant call to String.format()

Change-Id: I8f5fc09469b56d73d3838e7bcfecfd21140429eb
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
9 years agoFS_Win32: Avoid an IOException on Windows if bash is not in PATH 25/47025/3
Sebastian Schuberth [Mon, 4 May 2015 09:45:22 +0000 (11:45 +0200)]
FS_Win32: Avoid an IOException on Windows if bash is not in PATH

Change-Id: I3145f74ecee9f5b368e7f4b9fd7cb906f407eff5
Signed-off-by: Sebastian Schuberth <sschuberth@gmail.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSkip logging stack trace on corrupt objects 95/47395/1
Shawn Pearce [Wed, 6 May 2015 23:02:27 +0000 (16:02 -0700)]
Skip logging stack trace on corrupt objects

Instead of dumping a full stack trace when a client sends an invalid
commit, record only a short line explaining the attempt:

  Cannot receive Invalid commit c0ff33...: invalid author into /tmp/jgit.git

The text alone is sufficient to explain the problem and the stack
trace does not lend any additional useful information. ObjectChecker
is quite clear about its rejection cases.

Change-Id: Ifc8cf06032489dc6431be1ba66101cf3d4299218

9 years agoAdd repository name to failures in HTTP server log 92/47392/1
Shawn Pearce [Wed, 6 May 2015 22:47:34 +0000 (15:47 -0700)]
Add repository name to failures in HTTP server log

If UploadPack or ReceivePack has an exception record an identifier
associated with the repository as part of the log message. This can
help the HTTP admin track down the offending repository and take
action to repair the root cause.

Change-Id: I58f22b33cdb40994f044a26fba9fe965b45be51d

9 years agoMerge branch 'stable-4.0' 74/47174/1
Matthias Sohn [Tue, 5 May 2015 12:47:46 +0000 (14:47 +0200)]
Merge branch 'stable-4.0'

* stable-4.0:
  Prepare post 4.0.0-m2 builds
  JGit v4.0.0.201505050340-m2
  Prepare post 4.0.0-m1 builds
  JGit v4.0.0.201503231230-m1

Change-Id: I2cf68424bb550816f6137d693e792ebffa11bf87
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare post 4.0.0-m2 builds 68/47168/1
Matthias Sohn [Tue, 5 May 2015 10:41:54 +0000 (12:41 +0200)]
Prepare post 4.0.0-m2 builds

Change-Id: I11c2d5120508526ed2671d62f4dd6a395c3b4749
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v4.0.0.201505050340-m2 25/47125/1 v4.0.0.201505050340-m2
Matthias Sohn [Tue, 5 May 2015 07:36:57 +0000 (09:36 +0200)]
JGit v4.0.0.201505050340-m2

Change-Id: I4a360a654d5d5e4200bf1e4817613f85755bb3ff
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'master' into stable-4.0 92/47092/1
Matthias Sohn [Mon, 4 May 2015 20:02:23 +0000 (22:02 +0200)]
Merge branch 'master' into stable-4.0

Change-Id: I962461630384b76e7f387f4e1c1248833fbc4673
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare post 4.0.0-m1 builds 91/47091/1
Matthias Sohn [Mon, 4 May 2015 19:59:17 +0000 (21:59 +0200)]
Prepare post 4.0.0-m1 builds

Change-Id: I2e6a90d74f774d576e7d3595344dccaa27a96027
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Revert "ResolveMerge only needs to visit differing TreeEntries""
Matthias Sohn [Mon, 4 May 2015 13:33:27 +0000 (09:33 -0400)]
Merge "Revert "ResolveMerge only needs to visit differing TreeEntries""

9 years agoFix possible AIOOB in DirCacheTree.contains() 17/47017/1
Christian Halstrick [Mon, 4 May 2015 09:00:57 +0000 (11:00 +0200)]
Fix possible AIOOB in DirCacheTree.contains()

When DirCacheTree.contains() is called and 'aOff' is greater than 'aLen'
an ArrayIndexOutOfBoundsException was thrown. This fix makes
DirCacheTree.contains() more robust and allows parsing such index files
without throwing AIOOB.

I couldn't create a test case leading to this situation but I have seen
such situations while inspecting Bug: 465393. It seems that such
situations are created on Windows when there are invalid pathes in the
index. There may be a not yet known bug leading to such situations in
combination with invalid pathes.

Bug: 465393
Change-Id: I6535d924a22cba9a05df0ccd7e6dc2c9ddc42375

9 years agoUse CBI eclipse-jarsigner-plugin 1.1.2-SNAPSHOT 04/47004/2
Matthias Sohn [Mon, 4 May 2015 06:56:28 +0000 (08:56 +0200)]
Use CBI eclipse-jarsigner-plugin 1.1.2-SNAPSHOT

CBI recently fixed a couple of resource leaks which probably caused
jar signing failures on Hudson (bug 464947). JGit builds were also
affected. Hence use version 1.1.2-SNAPSHOT until a new release is
available.

Bug:464947
Change-Id: I7fb4a65f888194f7209c866cd58551891c89fb7a
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSupport agent= capability in wire protocol 28/46928/5
Shawn Pearce [Thu, 30 Apr 2015 19:43:46 +0000 (12:43 -0700)]
Support agent= capability in wire protocol

Since git-core ff5effd (v1.7.12.1) the native wire protocol transmits
the server and client implementation and version strings using
capability "agent=git/1.7.12.1" or similar.

Support this in JGit and hang the implementation data off UploadPack
and ReceivePack.  On HTTP transports default to the User-Agent HTTP
header until the client overrides this with the optional capability
string in the first line.

Extract the user agent string into a UserAgent class under transport
where it can be specified to a different value if the application's
build process has broken the Implementation-Version header in the
JGit package.

Change-Id: Icfc6524d84a787386d1786310b421b2f92ae9e65

9 years agoMerge changes I368af61a,I2400fe5f
Matthias Sohn [Thu, 30 Apr 2015 14:43:46 +0000 (10:43 -0400)]
Merge changes I368af61a,I2400fe5f

* changes:
  Remove pack from list when file handle is stale
  Lower log level to warn for handled pack errors

9 years agoAdd fsck.allowInvalidPersonIdent to accept invalid author/committers 35/46835/3
Shawn Pearce [Thu, 30 Apr 2015 04:56:28 +0000 (21:56 -0700)]
Add fsck.allowInvalidPersonIdent to accept invalid author/committers

A larger than expected number of real-world repositories found on
the Internet contain invalid author, committer and tagger lines
in their history. Many of these seem to be caused by users misusing
the user.name and user.email fields, e.g.:

  [user]
    name = Au Thor <author@example.com>
    email = author@example.com

that some version of Git (or a reimplementation thereof) copied
directly into the object header. These headers are not valid and
are rejected by a strict fsck, making it impossible to transfer
the repository with JGit/EGit.

Another form is an invalid committer line with double negative for
the time zone, e.g.

  committer Au Thor <a@b> 1288373970 --700

The real world is messy. :(

Allow callers and users to weaken the fsck settings to accept these
sorts of breakages if they really want to work on a repo that has
broken history.  Most routines will still function fine, however
commit timestamp sorting in RevWalk may become confused by a corrupt
committer line and sort commits out of order. This is mostly fine if
the corrupted chain is shorter than the slop window.

Change-Id: I6d529542c765c131de590f4f7ef8e7c1c8cb9db9

9 years agoRemove pack from list when file handle is stale 49/45649/5
Hugo Arès [Tue, 7 Apr 2015 15:17:06 +0000 (11:17 -0400)]
Remove pack from list when file handle is stale

This error happens on nfs file system when you try to read a file that
was deleted or replaced.

When the error happens because the file was deleted, removing it from
the list is the proper way to handle the error, same use case as
FileNotFoundException. When the error happens because the file was
replaced, removing the file from the list will cause the file to be
re-read so it will get the latest version of the file.

Bug: 462868
Change-Id: I368af61a6cf73706601a3e4df4ef24f0aa0465c5
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
9 years agoLower log level to warn for handled pack errors 41/45641/4
Hugo Arès [Fri, 10 Apr 2015 13:08:07 +0000 (09:08 -0400)]
Lower log level to warn for handled pack errors

Pack not found and pack corrupted/invalid are handled by the code (pack
is removed from the list) so logging an error and the stacktrace is
misleading because it implies that there is an action to take to fix the
error.

Lower the log level to warn and remove the stacktrace for those 2 types
of errors and keep the error log statement for any other.

Change-Id: I2400fe5fec07ac6d6c244b852cce615663774e6e
Signed-off-by: Hugo Arès <hugo.ares@ericsson.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'stable-3.7' 97/46597/1
Matthias Sohn [Mon, 27 Apr 2015 21:59:26 +0000 (23:59 +0200)]
Merge branch 'stable-3.7'

* stable-3.7:
  Prepare 3.7.2-SNAPSHOT builds
  JGit v3.7.1.201504261725-r
  Revert "Let ObjectWalk.markUninteresting also mark the root tree as"

Change-Id: If1b62ff695e063d797c3d13c43e488ca56f29cbe
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare 3.7.2-SNAPSHOT builds 23/46523/1 stable-3.7
Matthias Sohn [Sun, 26 Apr 2015 22:33:14 +0000 (00:33 +0200)]
Prepare 3.7.2-SNAPSHOT builds

Change-Id: Ia42a665e76953b643a080b58d2cd4526575ddf2e
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>