The Iff2de881 tried to fix missing tree ..." but introduced severe
performance degradation (>10x in some cases) when acting as server
(git push) and as client (replication). IOW cure is worse than the
disease.
This reverts commit c4797fe986.
Change-Id: I4e6056eb352d51277867f857a0cab380eca153ac
Signed-off-by: David Ostrovsky <david@ostrovsky.org>
Cached packs are only used when writing over the network or to
a bundle file and reuse validation is always disabled in these
two contexts. The client/consumer of the stream will be SHA-1
checksumming every object.
Reuse validation is most critical during local GC to avoid silently
ignoring corruption by stopping as soon as a problem is found and
leaving everything alone for the end-user to debug and salvage.
Cached packs are not supported during local GC as the bitmap rebuild
logic does not support including a cached pack in the result.
Strip out the validation and force PackWriter to always disable the
cached pack feature if reuseValidation is enabled.
Change-Id: If0d7baf2ae1bf1f7e71bf773151302c9f7887039
Sensible suggestion from Terry Parker as a late comment on
commit f2efcdc6f769d59722b17e9274932d585035cfb6.
Change-Id: I225775bfb6d3d91ae066ff00f9d80a9c02a422c2
This hint allows an underlying implementation to read more bytes when
possible and buffer them locally for future read calls to consume.
Change-Id: Ia986a1bb8640eecb91cfbd515c61fa1ff1574a6f
When a large pack (> 30% of the block cache) is being reused by
copying it pollutes the block cache with noise by storing blocks
that are never referenced again.
Avoid this by streaming the file directly from its channel onto
the output stream.
Change-Id: I2e53de27f3dcfb93de68b1fad45f75ab23e79fe7
The clone or fetch depth is a valuable bit of information
for access logging. Create a public getter to faciliate access.
A precondition check prevents unintentional misuse when the
data isn't valid yet.
Change-Id: I4603d5fd3bd4a767e3e2419b0f2da3664cfbd7f8
Signed-off-by: David Pletcher <dpletcher@google.com>
JGit hit IllegalArgumentException: invalid content length
when pushing large packs to S3.
Bug: 463015
Change-Id: Iddf50d90c7e3ccb15b9ff71233338c6b204b3648
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
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>
- close RevWalk allocated in scan()
- replace use of deprecated ObjectReader.release() method
Change-Id: I41b2b10a1a44270a6ceaa1741e996c0921439852
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Replaces use of deprecated release() methods.
Change-Id: I0211bcf0a76a2fccc2c85fa74778e20c256984ba
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
- replaces use of deprecated ObjectInserter.release()
- auto-close TreeWalk
Change-Id: I540ee711b8c3430a71fdff07add506b7d9c039dc
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
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
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
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
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
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>
This unintentionally was changed from severity debug to error which is
causing unexpected log entries.
Bug: 463349
Change-Id: I4b6d42a1420652ab6824e237bd231ba86896acbf
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
Bug: 390833
Change-Id: I29f7b79b241929877c93ac485c677487a91bb77b
Signed-off-by: André de Oliveira <andre.oliveira@liferay.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
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>
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
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>
This reverts commit 6bc48cdc62.
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>
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
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
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
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>
David Ostrovsky