]> source.dussan.org Git - jgit.git/log
jgit.git
9 years agoRemove "experimental" from the description of the pack bitmap index 57/52257/1
Terry Parker [Mon, 20 Jul 2015 18:28:11 +0000 (11:28 -0700)]
Remove "experimental" from the description of the pack bitmap index

Change-Id: I940701093eb6c6d26e147b79de77997b0df481a0
Signed-off-by: Terry Parker <tparker@google.com>
9 years agoDFS: Remove now-redundant getOneRef 16/52116/1
Jonathan Nieder [Thu, 16 Jul 2015 21:50:42 +0000 (14:50 -0700)]
DFS: Remove now-redundant getOneRef

Instead of reimplementing exactRef in a separate, private method,
use it directly.

Change-Id: I0fc06fc46eef0e36e571a6ef622f38dc2aa59038

9 years agoDFS: Override exactRef() with optimal version 15/52115/1
Shawn Pearce [Thu, 16 Jul 2015 21:31:50 +0000 (14:31 -0700)]
DFS: Override exactRef() with optimal version

This avoids scanning a directory to find all names and
then weeding down to the requested name.

Change-Id: I36ce31febad6e881182bc3a278c7ed35cc04cbda

9 years agoPushCertificate: Add toTextWithSignature() method 23/52023/2
Dave Borowitz [Thu, 16 Jul 2015 01:07:06 +0000 (18:07 -0700)]
PushCertificate: Add toTextWithSignature() method

Change-Id: Ic0129373cc0c0321ffa43dc11a658d4f891ec2c2

9 years agoPushCertificateParser: Add fromString method 22/52022/1
Dave Borowitz [Thu, 16 Jul 2015 01:04:10 +0000 (18:04 -0700)]
PushCertificateParser: Add fromString method

Change-Id: I74c3f65a9ff297f708d996a4c138456a31a466b8

9 years agoBaseReceivePack: Add method to override the push cert 21/52021/1
Dave Borowitz [Thu, 16 Jul 2015 00:58:26 +0000 (17:58 -0700)]
BaseReceivePack: Add method to override the push cert

Change-Id: Id912f45c933c78a85b245fa7b6edebbeb40b009c

9 years agoBaseReceivePack: Don't throw from getPushCertificate() 20/52020/1
Dave Borowitz [Thu, 16 Jul 2015 00:55:17 +0000 (17:55 -0700)]
BaseReceivePack: Don't throw from getPushCertificate()

Rather than lazily parsing the push in this method, parse it at the
end of recvCommands(), which already contains the necessary try/catch
for handling this error. This allows later callers to avoid having to
handle this condition superfluously.

Change-Id: I5dcaf1a44bf4e321adc281e3381e7e17ac89db06

9 years agoPushCertificateStore: Return boolean from batch save methods 16/52016/2
Dave Borowitz [Wed, 15 Jul 2015 17:18:01 +0000 (10:18 -0700)]
PushCertificateStore: Return boolean from batch save methods

Change-Id: I9730cb4f60c60ee6d5a7a156a0b6a53f79309ec3

9 years agoAllow saving push certs on a subset of refs 74/51874/4
Dave Borowitz [Mon, 13 Jul 2015 19:03:42 +0000 (12:03 -0700)]
Allow saving push certs on a subset of refs

Consider a BatchRefUpdate produced by Gerrit Code Review, where the
original command pushed over the wire might refer to
"refs/for/master", but that command is ignored and replaced with some
additional commands like creating "refs/changes/34/1234/1". We do not
want to store the cert in "refs/for/master@{cert}", since that may
lead someone looking to the ref to the incorrect conclusion that that
ref exists.

Add a separate put method that takes a collection of commands, and
only stores certs on those refs that have a matching command in the
cert.

Change-Id: I4661bfe2ead28a2883b33a4e3dfe579b3157d68a

9 years agoUpdate org.apache.httpcomponents 94/51894/1
Matthias Sohn [Mon, 13 Jul 2015 23:55:58 +0000 (01:55 +0200)]
Update org.apache.httpcomponents

- update org.apache.httpcomponents.httpcore to 4.3.3
- update org.apache.httpcomponents.httpclient to 4.3.6, 4.3.5 and later
  are reported to fix vulnerability CVE-2014-3577

CQ: 9220
CQ: 9221
Bug: 470523
Change-Id: I39b80b250c6c1daede6a23e9f177fb2988ac37bb
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAccess static member LocalDiskRepositoryTestCase.CONTENT directly 60/51860/2
Matthias Sohn [Mon, 13 Jul 2015 15:06:54 +0000 (17:06 +0200)]
Access static member LocalDiskRepositoryTestCase.CONTENT directly

37a1e4be moved this constant causing the following error message in
Eclipse: "The static field LocalDiskRepositoryTestCase.CONTENT should be
accessed directly".

Change-Id: I4ceb57a30f2e5a8f7e55109ef260a244ed5e7044
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPushCertificateStore: Add method to save in batch 66/51866/2
Dave Borowitz [Sat, 11 Jul 2015 00:12:54 +0000 (17:12 -0700)]
PushCertificateStore: Add method to save in batch

Change-Id: I8bfaee1a52d368ffe2cd7e8af1754a5261569078

9 years agoFixed @since tags on new API 01/51801/1
Andrey Loskutov [Sun, 12 Jul 2015 16:44:08 +0000 (18:44 +0200)]
Fixed @since tags on new API

Change-Id: I77bc2ec486ffe4c8b864663d103d8cf39c3f35e1
Signed-off-by: Andrey Loskutov <loskutov@gmx.de>
9 years agoStore push certificates in refs/meta/push-certs 28/51128/13
Dave Borowitz [Tue, 30 Jun 2015 01:05:11 +0000 (18:05 -0700)]
Store push certificates in refs/meta/push-certs

Inspired by a proposal from gitolite[1], where we store a file in
a tree for each ref name, and the contents of the file is the latest
push cert to affect that ref.

The main modification from that proposal (other than lacking the
out-of-git batching) is to append "@{cert}" to filenames, which allows
storing certificates for both refs/foo and refs/foo/bar. Those
refnames cannot coexist at the same time in a repository, but we do
not want to discard the push certificate responsible for deleting the
ref, which we would have to do if refs/foo in the push cert tree
changed from a tree to a blob.

The "@{cert}" syntax is at least somewhat consistent with
gitrevisions(7) wherein @{...} describe operators on ref names.

As we cannot (currently) atomically update the push cert ref with the
refs that were updated, this operation is inherently racy. Kick the can
down the road by pushing this burden on callers.

[1] https://github.com/sitaramc/gitolite/blob/cf062b8bb6b21a52f7c5002d33fbc950762c1aa7/contrib/hooks/repo-specific/save-push-signatures

Change-Id: Id3eb32416f969fba4b5e4d9c4b47053c564b0ccd

9 years agoAdd setTargetBranch in RepoCommand. 96/51696/3
Yuxuan 'fishy' Wang [Fri, 10 Jul 2015 05:12:28 +0000 (22:12 -0700)]
Add setTargetBranch in RepoCommand.

This will allow us to write the super project in a branch other than
master.

Change-Id: I578ed9ecbc6423416239e31ad644531dae9fb5c3
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoMove `RepositoryTestCase.indexState` to parent class 15/51615/2
Chris Price [Tue, 7 Jul 2015 12:06:50 +0000 (13:06 +0100)]
Move `RepositoryTestCase.indexState` to parent class

The test helper method `indexState` in `RepositoryTestCase` is
very useful for writing tests, even in cases where we need to
do things like create more than one repository for a test and
thus we don't want to use the built-in `db` member variable that
exists in `RepositoryTestCase`.  Since the method is static,
we can move it up to the parent class `LocalDiskRepositoryTestCase`,
where it can be used by tests that aren't a great fit for inheriting
directly from `RepositoryTestCase`.

Bug: 436200
Change-Id: I2b6de75c001d2d77ddb607488af246548784a67f
Signed-off-by: Chris Price <chris@puppetlabs.com>
9 years agoChange `indexState` to static method 14/51614/2
Chris Price [Tue, 7 Jul 2015 11:43:56 +0000 (12:43 +0100)]
Change `indexState` to static method

The `indexState` method is no longer referring to any
member variables from the class, so it can be made static.

Bug: 436200
Change-Id: I013316de5c373417ea758ca6e17da29209fead53
Signed-off-by: Chris Price <chris@puppetlabs.com>
9 years agoMerge "Use local variable in RepositoryTestCase.indexState"
Christian Halstrick [Thu, 9 Jul 2015 22:27:38 +0000 (18:27 -0400)]
Merge "Use local variable in RepositoryTestCase.indexState"

9 years agoPushCertificateParser: Move reader parsing to instance method 30/51630/3
Dave Borowitz [Thu, 9 Jul 2015 01:42:28 +0000 (18:42 -0700)]
PushCertificateParser: Move reader parsing to instance method

Change-Id: I8ab5fb97303e25b28187375b6f9ff28f918bdd43

9 years agoPushCertificateParser: Make constructor public 29/51629/3
Dave Borowitz [Thu, 9 Jul 2015 01:41:32 +0000 (18:41 -0700)]
PushCertificateParser: Make constructor public

Primarily to aid in writing end-to-end tests, which must be written
outside this package as we don't currently have a Bouncy Castle
dependency.

Change-Id: Ia24179773140d9ecc616b3a6b2e630c85d800157

9 years agoReport PackProtocolExceptions to client during receive-pack 09/51209/6
Dave Borowitz [Wed, 1 Jul 2015 18:24:25 +0000 (11:24 -0700)]
Report PackProtocolExceptions to client during receive-pack

We have done this since forever with the "wanted old new ref" error,
so let's do it for other such errors thrown in the same block as well.

Change-Id: Ib3b1c7f05e31a5b3e40e85eb07b16736920a033b

9 years agoPushCertificate: implement hashCode/equals/toString 04/51204/7
Dave Borowitz [Wed, 1 Jul 2015 16:42:35 +0000 (09:42 -0700)]
PushCertificate: implement hashCode/equals/toString

Change-Id: Ib588a3f47492cee5e5e6274a3b088678919a0fa0

9 years agoPushCertificateParser: Make pushee optional 08/51208/9
Dave Borowitz [Wed, 1 Jul 2015 18:44:09 +0000 (11:44 -0700)]
PushCertificateParser: Make pushee optional

When pushing to an HTTP server using the C git client, I observed a
certificate lacking a pushee field. Handle this gracefully in the
parser.

Change-Id: I7f3c5fa78f2e35172a93180036e679687415cac4

9 years agoPushCertificateParser: Change error string for invalid field 07/51207/8
Dave Borowitz [Wed, 1 Jul 2015 18:29:09 +0000 (11:29 -0700)]
PushCertificateParser: Change error string for invalid field

The old one didn't take an argument; the new one includes the field
name.

Change-Id: I175dc75dcdd05ba6fdec315c04c79550cfdc2342

9 years agoPushCertificateParser: Add method for parsing from a stream 27/51127/11
Dave Borowitz [Tue, 30 Jun 2015 00:26:57 +0000 (17:26 -0700)]
PushCertificateParser: Add method for parsing from a stream

We intend to store received push certificates somewhere, like a
particular ref in the repository in question. For reading data back
out, it will be useful to read push certificates (without pkt-line
framing) in a streaming fashion.

Change-Id: I70de313b1ae463407b69505caee63e8f4e057ed4

9 years agoIO: Add a method for reading lines 26/51126/8
Dave Borowitz [Tue, 30 Jun 2015 00:26:39 +0000 (17:26 -0700)]
IO: Add a method for reading lines

Change-Id: Ib7be76aa7ac889354ad4782e2b64d4221a0e25b9

9 years agoBaseReceivePack: Treat all LFs as optional 36/51436/3
Dave Borowitz [Mon, 6 Jul 2015 19:19:42 +0000 (15:19 -0400)]
BaseReceivePack: Treat all LFs as optional

Discussion on the git mailing list has concluded[1] that the intended
behavior for all (non-sideband) portions of the receive-pack protocol
is for trailing LFs in pkt-lines to be optional. Go back to using
PacketLineIn#readString() everywhere.

For push certificates specifically, we agreed that the payload signed
by the client is always concatenated with LFs even though the client
MAY omit LFs when framing the certificate for the wire. This is still
reflected in the implementation of PushCertificate#toText().

[1] http://thread.gmane.org/gmane.comp.version-control.git/273175/focus=273412

Change-Id: I817231c4d4defececb8722142fea18ff42e06e44

9 years agoBaseReceivePack: More validation during parseCommand 73/51073/5
Dave Borowitz [Mon, 29 Jun 2015 23:00:53 +0000 (16:00 -0700)]
BaseReceivePack: More validation during parseCommand

Change-Id: I25f3a5582a45dd0ec8f78f5daf74c2203797a184

9 years agoUse local variable in RepositoryTestCase.indexState 75/51475/1
Chris Price [Tue, 7 Jul 2015 11:30:06 +0000 (12:30 +0100)]
Use local variable in RepositoryTestCase.indexState

There is a signature of the test helper method `indexState`,
in `RepositoryTestCase`, that accepts a `Repository` object
as an argument.  However, there was one line of code where
this variable was not being used, and the method was instead
referring to a member variable `db`.  I believe this was
probably just an oversight in a previous refactor, and
that the correct behavior is to use the variable from
the argument list.  This change also has the benefit
of making it possible to convert this method to a static
method, since it no longer relies on any state from the class.

Bug: 436200
Change-Id: Iac95b046dc5bd0b3756642e241c3637f1fad3609
Signed-off-by: Chris Price <chris@puppetlabs.com>
9 years agoFix non-escaped quotes in JGitText.properties 66/51366/3
Matthias Sohn [Fri, 3 Jul 2015 23:08:59 +0000 (01:08 +0200)]
Fix non-escaped quotes in JGitText.properties

In most texts we use "cannot" hence instead of escaping the apostroph in
"can't" use "cannot".

Bug: 471796
Change-Id: Icda5b4db38076789d06498428909306aef3cb68b
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoThrow InvalidObjectIdException from ObjectId.fromString("tooshort") 44/51144/3
Jonathan Nieder [Tue, 30 Jun 2015 21:42:39 +0000 (14:42 -0700)]
Throw InvalidObjectIdException from ObjectId.fromString("tooshort")

ObjectId.fromString already throws InvalidObjectIdException for most
malformed object ids, but for this kind it previously threw
IllegalArgumentException.  Since InvalidObjectIdException is a child of
IllegalArgumentException, callers that catch IllegalArgumentException
will continue to work.

Change-Id: I24e1422d51607c86a1cb816a495703279e461f01
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoAvoid double-colon in InvalidObjectIdException description 43/51143/2
Jonathan Nieder [Tue, 30 Jun 2015 21:34:02 +0000 (14:34 -0700)]
Avoid double-colon in InvalidObjectIdException description

The invalidId message in JGitText and the asAscii bad id both contain a
colon, so the resulting message would say

Invalid id: : a78987c98798ufa

Fix it by keeping the colon in the translated message and not adding
another colon programmatically.

Noticed by code inspection.

Change-Id: I13972eebde27a4128828e6c64517666f0ba6288b
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoCompare API against 4.0.0.201506090130-r 09/50809/1
Matthias Sohn [Wed, 24 Jun 2015 13:14:54 +0000 (15:14 +0200)]
Compare API against 4.0.0.201506090130-r

Change-Id: I7cb7e89875b5238a54411b8c43c298d4cbe9f7ac
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge branch 'stable-4.0' 95/50795/1
Matthias Sohn [Wed, 24 Jun 2015 12:46:51 +0000 (14:46 +0200)]
Merge branch 'stable-4.0'

Change-Id: I5c965206ad10bababe366a51ab7c33a8836a7868
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare 4.0.2-SNAPSHOT builds 84/50784/1
Matthias Sohn [Wed, 24 Jun 2015 11:51:21 +0000 (13:51 +0200)]
Prepare 4.0.2-SNAPSHOT builds

Change-Id: I645cacfdde21aa28aa2e17c10dec0576b170ed0e
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v4.0.1.201506240215-r 61/50761/1 v4.0.1.201506240215-r
Matthias Sohn [Wed, 24 Jun 2015 06:17:29 +0000 (08:17 +0200)]
JGit v4.0.1.201506240215-r

Change-Id: Ib7713b657e7812b0debd72bb4eece0daa187e80d
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAPI to remove repositories from RepositoryCache 38/50238/3
Tobias Oberlies [Tue, 16 Jun 2015 08:14:44 +0000 (10:14 +0200)]
API to remove repositories from RepositoryCache

Add methods that allow to unregister repositories from the
RepositoryCache individually.

Bug: 470234
Change-Id: Ib918a634d829c9898072ae7bdeb22b099a32b1c9
Signed-off-by: Tobias Oberlies <tobias.oberlies@sap.com>
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd a grace period for packfiles during GC 30/50230/2
Christian Halstrick [Fri, 12 Jun 2015 23:38:29 +0000 (01:38 +0200)]
Add a grace period for packfiles during GC

For loose objects an expiration date can be set which will save too
young objects from being deleted. Add the same for packfiles. Packfiles
which are too young are not deleted.

Bug: 468024
Change-Id: I3956411d19b47aaadc215dab360d57fa6c24635e
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoSignedPushConfig: Allow setting a custom nonce generator impl 61/50461/1
Dave Borowitz [Thu, 18 Jun 2015 17:07:37 +0000 (13:07 -0400)]
SignedPushConfig: Allow setting a custom nonce generator impl

Change-Id: Ic0156a7d65d99881ef27801fcce7754594c436f0

9 years agoPackCertificateParser: return null if nothing was received 51/50451/2
Dave Borowitz [Thu, 18 Jun 2015 14:55:16 +0000 (10:55 -0400)]
PackCertificateParser: return null if nothing was received

Add test for this case in both the enabled and disabled cases.

Change-Id: If9d12192a2dc9f9dd1eac9844b5a7b0edadc0b34

9 years agoInclude push certificates in RefUpdate 47/50447/2
Dave Borowitz [Thu, 18 Jun 2015 14:14:48 +0000 (10:14 -0400)]
Include push certificates in RefUpdate

This may be used by e.g. a custom reflog implementation to record
this information along with the ref update.

Change-Id: I44adbfad704b76f9c1beced6e1ce82eaf71410d2

9 years agoRemove redundant @since tags in new signed push classes 46/50446/1
Dave Borowitz [Thu, 18 Jun 2015 14:18:38 +0000 (10:18 -0400)]
Remove redundant @since tags in new signed push classes

Change-Id: I9f57318a59f668906b80f67230d36ba6439fc7b4

9 years agoAdd a separate type for the identity in a push certificate 94/50394/4
Dave Borowitz [Wed, 17 Jun 2015 17:12:22 +0000 (13:12 -0400)]
Add a separate type for the identity in a push certificate

These differ subtly from a PersonIdent, because they can contain
anything that is a valid User ID passed to gpg --local-user. Upstream
git push --signed will just take the configuration value from
user.signingkey and pass that verbatim in both --local-user and the
pusher field of the certificate. This does not necessarily contain an
email address, which means the parsing implementation ends up being
substantially different from RawParseUtils.parsePersonIdent.
Nonetheless, we try hard to match PersonIdent behavior in
questionable cases.

Change-Id: I37714ce7372ccf554b24ddbff56aa61f0b19cbae

9 years agoPushCertificateParser: include begin/end lines in signature 96/50296/6
Dave Borowitz [Mon, 15 Jun 2015 20:50:22 +0000 (16:50 -0400)]
PushCertificateParser: include begin/end lines in signature

The signature is intended to be passed to a verification library such
as Bouncy Castle, which expects these lines to be present in order to
parse the signature.

Change-Id: I22097bead2746da5fc53419f79761cafd5c31c3b

9 years agoPushCertificateParser: throw PackProtocolException in more cases 95/50295/6
Dave Borowitz [Mon, 15 Jun 2015 19:48:22 +0000 (15:48 -0400)]
PushCertificateParser: throw PackProtocolException in more cases

This is the subclass of IOException already thrown by
BaseReceivePack#recvCommands when encountering an invalid value on
the wire. That's what PushCertificateParser is doing too, so use the
same subclass.

Change-Id: I1d323909ffe70757ea56e511556080695b1a0c11

9 years agoExtract a class for signed push configuration 94/50294/6
Dave Borowitz [Mon, 15 Jun 2015 19:25:14 +0000 (15:25 -0400)]
Extract a class for signed push configuration

The default behavior is to read a repository's signed push
configuration from that repo's config file, but this is not very
flexible when it comes to managing groups of repositories (e.g. with
Gerrit). Allow callers to override the configuration using a POJO.

Change-Id: Ib8f33e75daa0b2fbd000a2c4558c01c014ab1ce5

9 years agoBaseReceivePack: fix reading cert lines in command loop 93/50293/4
Dave Borowitz [Mon, 15 Jun 2015 18:54:13 +0000 (14:54 -0400)]
BaseReceivePack: fix reading cert lines in command loop

Add a missing continues to prevent falling through to the command
parsing section. The first continue happens when the command list is
empty, so change the condition to see whether we have read the first
line, rather than any commands.

Fix comparison to BEGIN_SIGNATURE to use raw line with newline.

Change-Id: If3d92f5ceade8ba7605847a4b2bc55ff17d119ac

9 years agoSilence API warnings introduced by a85e817d 95/50395/1
Matthias Sohn [Wed, 17 Jun 2015 21:08:45 +0000 (23:08 +0200)]
Silence API warnings introduced by a85e817d

a85e817d is a slightly breaking API change to classes that were
technically public and technically released in 4.0. However, it is
highly unlikely that people were actually depending on public behavior,
since there were no public methods to create PushCertificates with
anything other than null field values, or a PushCertificateParser that
did anything other than infinite loop or throw exceptions when reading.

Change-Id: I1d0ba9ea0a347e8ff5a0f4af169d9bb18c5838d2
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoUpdate comments about PackStatistics 91/50291/1
Terry Parker [Tue, 16 Jun 2015 17:05:29 +0000 (10:05 -0700)]
Update comments about PackStatistics

Change-Id: I4862932dc38b58cc98400ed584cc506013846065
Signed-off-by: Terry Parker <tparker@google.com>
9 years agoAdd the set of shallow commits to PackStatistics 21/50221/2
Terry Parker [Mon, 15 Jun 2015 21:45:12 +0000 (14:45 -0700)]
Add the set of shallow commits to PackStatistics

Change-Id: I976102e054369b045340d4b9209984b5243e3750
Signed-off-by: Terry Parker <tparker@google.com>
9 years agoAdd tests for HMACSHA1NonceGenerator 83/50183/2
Dave Borowitz [Mon, 15 Jun 2015 14:14:59 +0000 (10:14 -0400)]
Add tests for HMACSHA1NonceGenerator

Correct documentation of NonceStatus.OK/SLOP to match the implemented
behavior.

Change-Id: Id5ec1945eab76db6d2e4b592cb25907ea3d835cd

9 years agoMerge "Introduce PostUploadHook to replace UploadPackLogger"
Shawn Pearce [Fri, 12 Jun 2015 21:21:38 +0000 (17:21 -0400)]
Merge "Introduce PostUploadHook to replace UploadPackLogger"

9 years agoIntroduce PostUploadHook to replace UploadPackLogger 30/50130/5
Terry Parker [Fri, 12 Jun 2015 19:00:36 +0000 (12:00 -0700)]
Introduce PostUploadHook to replace UploadPackLogger

UploadPackLogger is incorrectly named--it can be used to trigger any
post upload action, such as GC/compaction. This change introduces
PostUploadHook/PostUploadHookChain to replace
UploadPackLogger/UploadPackLoggerChain and deprecates the latter.

It also introduces PackStatistics as a replacement for
PackWriter.Statistics, since the latter is not public API.
It changes PackWriter to use PackStatistics and reimplements
PackWriter.Statistics to delegate to PackStatistics.

Change-Id: Ic51df1613e471f568ffee25ae67e118425b38986
Signed-off-by: Terry Parker <tparker@google.com>
9 years agoTreat CloneCommand.setBranch(null) as setBranch("HEAD") 78/49978/2
Jonathan Nieder [Wed, 10 Jun 2015 22:43:48 +0000 (15:43 -0700)]
Treat CloneCommand.setBranch(null) as setBranch("HEAD")

This method is documented to take a branch name (not a possibly null
string).  The only way a caller could have set null without either
re-setting to a sane value afterward or producing NullPointerException
was to also call setNoCheckout(true), in which case there would have
been no reason to set the branch in the first place.

Make setBranch(null) request the default behavior (remote's default
branch) instead, imitating C git's clone --no-branch.

Change-Id: I960e7046b8d5b5bc75c7f3688f3a075d3a951b00
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoTreat CloneCommand.setRemote(null) as setRemote("origin") 77/49977/2
Jonathan Nieder [Wed, 10 Jun 2015 22:43:27 +0000 (15:43 -0700)]
Treat CloneCommand.setRemote(null) as setRemote("origin")

A non-bare clone command with null remote produces a
NullPointerException when trying to produce a refspec to fetch against.

In a bare repository, a null remote name is accepted by mistake,
producing a configuration with items like 'remote.url' instead of
'remote.<remote>.url'.  This was never meant to work.

Instead, let's make setRemote(null) undo any previous setRemote calls
and re-set the remote name to DEFAULT_REMOTE, imitating C git clone's
--no-origin option.

While we're here, add some tests for setRemote working normally.

Change-Id: I76f502da5e677df501d3ef387e7f61f42a7ca238
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoDocument CloneCommand.setURI(null) 76/49976/2
Jonathan Nieder [Wed, 10 Jun 2015 22:11:39 +0000 (15:11 -0700)]
Document CloneCommand.setURI(null)

call() throws InvalidRemoteException if uri == null, so there should
never be reason to leave the URI set to null.  Document this.

Change-Id: I7f2cdbe8042d99cf8a3c1a8c4c2dcb58c5b8c305
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoHandle null in ProgressMonitor setters 80/49980/2
Jonathan Nieder [Wed, 10 Jun 2015 20:59:48 +0000 (13:59 -0700)]
Handle null in ProgressMonitor setters

These commands' monitor fields can never be null unless someone passes
null to setProgressMonitor.  Anyone passing null probably meant to
disable the ProgressMonitor, so do that (by falling back to
NullProgressMonitor.INSTANCE) instead of saving a null and eventually
producing NullPointerException.

Change-Id: I63ad93ea8ad669fd333a5fd40880e7583ba24827
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoDocument CloneCommand.setDirectory(null) and setGitDir(null) more clearly 79/49979/2
Jonathan Nieder [Wed, 10 Jun 2015 22:52:34 +0000 (15:52 -0700)]
Document CloneCommand.setDirectory(null) and setGitDir(null) more clearly

Change-Id: I74757ca68eaa941833846a7c8d983b4e996fa207
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoRequire all projects to have a name 81/49981/3
Jonathan Nieder [Wed, 10 Jun 2015 23:35:43 +0000 (16:35 -0700)]
Require all projects to have a name

A project's name attribute is required according to repo's
doc/manifest-format.txt:

    <!ELEMENT project (annotation*,
                       project*)>
    <!ATTLIST project name        CDATA #REQUIRED>

Enforcing this in code makes reading easier (in particular making it
clearer that getName() and getPath() can never return null).

Change-Id: I8c7014dd6042183d7fecb2202af5acdc384aa8e4
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge topic 'push-cert-2'
Dave Borowitz [Thu, 11 Jun 2015 17:07:42 +0000 (13:07 -0400)]
Merge topic 'push-cert-2'

* changes:
  Rewrite push certificate parsing
  Allow trailing newlines in receive-pack

9 years agoRewrite push certificate parsing 66/49966/7
Dave Borowitz [Wed, 10 Jun 2015 00:23:03 +0000 (17:23 -0700)]
Rewrite push certificate parsing

- Consistently return structured data, such as actual ReceiveCommands,
  which is more useful for callers that are doing things other than
  verifying the signature, e.g. recording the set of commands.
- Store the certificate version field, as this is required to be part
  of the signed payload.
- Add a toText() method to recreate the actual payload for signature
  verification. This requires keeping track of the un-chomped command
  strings from the original protocol stream.
- Separate the parser from the certificate itself, so the actual
  PushCertificate object can be immutable. Make a fair attempt at deep
  immutability, but this is not possible with the current mutable
  ReceiveCommand structure.
- Use more detailed error messages that don't involve NON-NLS strings.
- Document null return values more thoroughly. Instead of having the
  undocumented behavior of throwing NPE from certain methods if they
  are not first guarded by enabled(), eliminate enabled() and return
  null from those methods.
- Add tests for parsing a push cert from a section of pkt-line stream
  using a real live stream captured with Wireshark (which, it should
  be noted, uncovered several simply incorrect statements in C git's
  Documentation/technical/pack-protocol.txt).

This is a slightly breaking API change to classes that were
technically public and technically released in 4.0. However, it is
highly unlikely that people were actually depending on public
behavior, since there were no public methods to create
PushCertificates with anything other than null field values, or a
PushCertificateParser that did anything other than infinite loop or
throw exceptions when reading.

Change-Id: I5382193347a8eb1811032d9b32af9651871372d0

9 years agoMerge "submodule test: Use config.unset instead of setting to null"
Christian Halstrick [Thu, 11 Jun 2015 11:44:33 +0000 (07:44 -0400)]
Merge "submodule test: Use config.unset instead of setting to null"

9 years agoMerge "Add new submodule layout to SubmoduleAddCommand"
Christian Halstrick [Thu, 11 Jun 2015 09:13:21 +0000 (05:13 -0400)]
Merge "Add new submodule layout to SubmoduleAddCommand"

9 years agoAllow trailing newlines in receive-pack 65/49965/5
Dave Borowitz [Wed, 10 Jun 2015 19:47:04 +0000 (12:47 -0700)]
Allow trailing newlines in receive-pack

C git's receive-pack.c strips trailing newlines in command lists when
present[1], although send-pack.c does not send them, at least in the
case of command lists[2]. Change JGit to match this behavior.
Add tests.

This also fixes parsing of commands in the push cert, which, unlike
commands sent in the non-push case, always have trailing newlines.

[1] https://github.com/git/git/blob/7974889a053574e449b55ca543a486e38e74864f/builtin/receive-pack.c#L1380
where packet_read_line chomps newlines:
https://github.com/git/git/blob/7974889a053574e449b55ca543a486e38e74864f/pkt-line.c#L202

[2] https://github.com/git/git/blob/7974889a053574e449b55ca543a486e38e74864f/send-pack.c#L470

Change-Id: I4bca6342a7482a53c9a5815a94b3c181a479d04b

9 years agoAdd new submodule layout to SubmoduleAddCommand 34/49734/2
Christian Halstrick [Tue, 9 Jun 2015 07:31:26 +0000 (09:31 +0200)]
Add new submodule layout to SubmoduleAddCommand

The new submodule layout where GITDIR of a submodule is located at
<parent-repo-GITDIR>/modules/<submodule-path> was only used during
clone. Teach SubmoduleAddCommand to use the new layout.

Bug: 469666
Change-Id: Ie97dc0607b71499560444616f362bccee9cce515
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoTweak exception Javadoc to be eclipse-clean 75/49975/1
Jonathan Nieder [Wed, 10 Jun 2015 22:08:41 +0000 (15:08 -0700)]
Tweak exception Javadoc to be eclipse-clean

Bug: 469887
Change-Id: Ic189efc50d9c1f573dda0c79e5b0dad0813a5577
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agosubmodule test: Use config.unset instead of setting to null 74/49974/1
Jonathan Nieder [Wed, 10 Jun 2015 21:48:21 +0000 (14:48 -0700)]
submodule test: Use config.unset instead of setting to null

Most relative-URL tests for SubmoduleInitCommand carry out the following
steps:

 1. add a submodule at path "sub" to the index
 2. set remote.origin.url in .git/config
 3. configure .gitmodules, possibly using relative URLs, and see what
    happens

resolveWorkingDirectoryRelativeUrl() is meant to test the fallback when
remote.origin.url is not set, to match C git which treats the URL as
relative to the cwd in that case.  To do so, in step (2) it sets
remote.origin.url to null.

However, Config.setString when taking a null value does not actually
unset that value from the configuration --- it sets it to the empty
string.  This means we are testing a behavior that C git never
supported.  Use Config.unset instead.

Change-Id: I7af29fbbd333a2598843d62c320093c48b2ad972
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge changes I38e1798e,Id1fe1c20,I945ba188,I2c6be9cb
Jonathan Nieder [Wed, 10 Jun 2015 20:40:10 +0000 (16:40 -0400)]
Merge changes I38e1798e,Id1fe1c20,I945ba188,I2c6be9cb

* changes:
  Allow setting detail message and cause when constructing most exceptions
  Use message from ServiceNotAuthorizedException, ServiceNotEnabledException
  dumb HTTP: Clarify AsIsFilter by introducing req and res locals
  Clarify description of ServiceNotAuthorizedException

9 years agoMerge "UploadPack: Use reachable-sha1-in-want configuration"
Shawn Pearce [Wed, 10 Jun 2015 02:14:35 +0000 (22:14 -0400)]
Merge "UploadPack: Use reachable-sha1-in-want configuration"

9 years agoAdd missing imports for javadoc to ObjectCountCallback 81/49881/1
Jonathan Nieder [Wed, 10 Jun 2015 01:36:58 +0000 (18:36 -0700)]
Add missing imports for javadoc to ObjectCountCallback

Change-Id: I24780a28c01047654d81af09045d88c6a84950b6
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge "Fix public API issues introduced in I1baeedcc6946."
Jonathan Nieder [Wed, 10 Jun 2015 00:45:15 +0000 (20:45 -0400)]
Merge "Fix public API issues introduced in I1baeedcc6946."

9 years agoFix public API issues introduced in I1baeedcc6946. 72/49872/3
Yuxuan 'fishy' Wang [Wed, 10 Jun 2015 00:07:55 +0000 (17:07 -0700)]
Fix public API issues introduced in I1baeedcc6946.

Move ObjectCountCallback and WriteAbortedException to package
org.eclipse.jgit.transport, so that they'll become public API.

Change-Id: I95e3cfaa49f3f7371e794d5c253cf6981f87cae0
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoDocument empty catch block to silence warning 68/49868/2
Matthias Sohn [Tue, 9 Jun 2015 23:39:52 +0000 (01:39 +0200)]
Document empty catch block to silence warning

Change-Id: I643f916e394ca95272b58d0b5826e81cc1092f2f
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Fix imports in BundleWriter"
Jonathan Nieder [Wed, 10 Jun 2015 00:12:05 +0000 (20:12 -0400)]
Merge "Fix imports in BundleWriter"

9 years agoFix imports in BundleWriter 66/49866/2
Matthias Sohn [Tue, 9 Jun 2015 23:27:19 +0000 (01:27 +0200)]
Fix imports in BundleWriter

Change-Id: I5eb186ad27c93feab2664fcbcc9ca0b617a15462
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoMerge "Add missing javadoc @param for new method setObjectCountCallback()"
Jonathan Nieder [Tue, 9 Jun 2015 23:47:02 +0000 (19:47 -0400)]
Merge "Add missing javadoc @param for new method setObjectCountCallback()"

9 years agoMerge "Revert "Config: Distinguish between empty and null strings""
Jonathan Nieder [Tue, 9 Jun 2015 23:45:33 +0000 (19:45 -0400)]
Merge "Revert "Config: Distinguish between empty and null strings""

9 years agoAdd missing javadoc @param for new method setObjectCountCallback() 65/49865/2
Matthias Sohn [Tue, 9 Jun 2015 23:24:03 +0000 (01:24 +0200)]
Add missing javadoc @param for new method setObjectCountCallback()

Change-Id: I91acdc3680bde09ae5809c14460500223a410395
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoRemove no longer used API warning filter for FileUtil 64/49864/1
Matthias Sohn [Tue, 9 Jun 2015 23:21:21 +0000 (01:21 +0200)]
Remove no longer used API warning filter for FileUtil

Change-Id: Ic13943139f23cdb1eac840417de43796236d578f
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoRevert "Config: Distinguish between empty and null strings" 62/49862/3
Jonathan Nieder [Tue, 9 Jun 2015 22:54:52 +0000 (15:54 -0700)]
Revert "Config: Distinguish between empty and null strings"

This reverts commit 96eb3ee3976e7e9e3e118851fa614cce8a1f7d88, which
broke Gerrit tests that set a config value to 'null', serialize the
result, deserialize, and expect 'null' from Config.getString[1].

The intent of that commit was to make it possible to distinguish between
an absent and an empty config value, which we'll have to do with a new
method.

Revert the behavior change.  Keep the tests from 428cb23f2de8, since
they test the behavior more precisely than the old tests did.

[1] https://gerrit-review.googlesource.com/68452

Change-Id: Ie8042f380ea0e34e3203e1991aa0feb2e6e44641
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoFix a javadoc typo in RepoProject. 56/49856/1
Yuxuan 'fishy' Wang [Tue, 9 Jun 2015 21:24:36 +0000 (14:24 -0700)]
Fix a javadoc typo in RepoProject.

Change-Id: I3ea25c157b9e11a46a140d2c3ab9373d1596c163
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoCallback in PackWriter & BundleWriter. 86/49486/6
Yuxuan 'fishy' Wang [Thu, 4 Jun 2015 23:43:24 +0000 (16:43 -0700)]
Callback in PackWriter & BundleWriter.

Added callback in PackWriter and BundleWriter for the caller to get the
count of objects to write, and a chance to abort the write operation.

Change-Id: I1baeedcc6946b1093652de4a707fe597a577e526
Signed-off-by: Yuxuan 'fishy' Wang <fishywang@google.com>
9 years agoUploadPack: Use reachable-sha1-in-want configuration 52/49652/2
Fredrik Medley [Fri, 29 May 2015 18:04:50 +0000 (20:04 +0200)]
UploadPack: Use reachable-sha1-in-want configuration

C git 2.5 supports setting the equivalent of
RequestPolicy.REACHABLE_COMMIT with uploadpack.allowreachablesha1inwant.
Parse this into TransportConfig and use it from UploadPack. An explicitly
set RequestPolicy overrides the config, and the policy may still be
upgraded on a unidirectional connection to avoid races.

Change-Id: Id39771a6e42d8082099acde11249306828a053c0
Signed-off-by: Fredrik Medley <fredrik.medley@gmail.com>
9 years agoMerge branch 'stable-4.0' 02/49802/2
Matthias Sohn [Tue, 9 Jun 2015 14:56:16 +0000 (16:56 +0200)]
Merge branch 'stable-4.0'

* stable-4.0:
  Prepare 4.0.1-SNAPSHOT builds
  JGit v4.0.0.201506090130-r
  Update build to use Tycho 0.23

Change-Id: I732b601c198ad1a50aa7ba3c1d1fb516b10912b5
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoPrepare 4.0.1-SNAPSHOT builds 87/49787/1
Matthias Sohn [Tue, 9 Jun 2015 13:17:22 +0000 (15:17 +0200)]
Prepare 4.0.1-SNAPSHOT builds

Change-Id: I51d03d1a47d1e3cd453701e397750749867028a2
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoJGit v4.0.0.201506090130-r 29/49729/1 v4.0.0.201506090130-r
Matthias Sohn [Tue, 9 Jun 2015 05:29:27 +0000 (07:29 +0200)]
JGit v4.0.0.201506090130-r

Change-Id: I01ad84fc74555656c42934cd62a85269a7030557
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAdd missing # to javadoc comment 90/49690/1
Jonathan Nieder [Mon, 8 Jun 2015 17:47:15 +0000 (10:47 -0700)]
Add missing # to javadoc comment

Change-Id: I351cee0ba2e77e3360846ac0c5368da3a322725c
Reported-by: Markus Keller <markus_keller@ch.ibm.com>
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoUpdate build to use Tycho 0.23 42/49642/1
Matthias Sohn [Mon, 8 Jun 2015 10:59:28 +0000 (12:59 +0200)]
Update build to use Tycho 0.23

see https://wiki.eclipse.org/Tycho/Release_Notes/0.23

Change-Id: I70e6711294e2a92efa77fd9dd1c906c449c9b4db
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoAllow lookup of multiple exact refs in one shot 77/49577/2
Jonathan Nieder [Fri, 5 Jun 2015 23:08:55 +0000 (16:08 -0700)]
Allow lookup of multiple exact refs in one shot

exactRef(ref1, ref2, ref3) requests multiple specific refs in a single
lookup, which may be faster in some backends than looking them up one by
one.

firstExactRef generalizes getRef by finding the first existing ref from
the list of refs named.  Its main purpose is for the default
implementation of getRef (finding the first existing ref in a search
path).  Hopefully it can be useful for other operations that look for
refs in a search path (e.g., git log --notes=<name>), too.

Change-Id: I5c6fcf1d3920f6968b8b97f3d4c3a267258c4b86
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoIntroduce exactRef to read a ref whose exact name is known 48/49548/4
Jonathan Nieder [Fri, 5 Jun 2015 21:14:55 +0000 (14:14 -0700)]
Introduce exactRef to read a ref whose exact name is known

Unlike getRef(name), the new exactRef method does not walk the search
path.  This should produce a less confusing result than getRef when the
exact ref name is known: it will not try to resolve refs/foo/bar to
refs/heads/refs/foo/bar even when refs/foo/bar does not exist.

It can be faster than both getRefs(ALL).get(name) and getRef(name)
because it only needs to examine a single ref.

A follow-up change will introduce a findRef synonym to getRef and
deprecate getRef to make the choice a caller is making more obvious
(exactRef or findRef, with the same semantics as getRefs(ALL).get and
getRefs(ALL).findRef).

Change-Id: If1bd09bcfc9919e7976a4d77f13184ea58dcda52
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoMerge "Restore checkObjectCollisions flag"
Shawn Pearce [Thu, 4 Jun 2015 20:35:50 +0000 (16:35 -0400)]
Merge "Restore checkObjectCollisions flag"

9 years agoRestore checkObjectCollisions flag 72/49472/2
David Pletcher [Thu, 4 Jun 2015 18:48:32 +0000 (11:48 -0700)]
Restore checkObjectCollisions flag

I am developing an offline pack verification feature based on
PackParser. The birthday collision check is a prohibitive obstacle
to performance at scale because it interacts with the repository
to perform collision checks. This CL restores the checkObjectCollisions
flag that was removed in 9638e0aa87614a6fb4f109bbeac0cde3462b9769,
while changing the flag getter and setter to protected from public as a
precaution against misuse.

Change-Id: I363cd0c9de57c5e8659cdfe2d51b17823f4fe793
Signed-off-by: David Pletcher <dpletcher@google.com>
9 years agoConfig: Allow ending a file with "=" and no newline 70/49470/1
Dave Borowitz [Thu, 4 Jun 2015 18:48:46 +0000 (11:48 -0700)]
Config: Allow ending a file with "=" and no newline

This is a perfectly valid construction according to C git:

$ echo -en '[a]\nx =' > foo.config
$ git config -f foo.config a.x; echo $?

0

Change-Id: Icfcf8304adb43c79e2b8b998f8d651b2a94f6acb

9 years agoConfig: Distinguish between empty and null strings 69/49469/1
Dave Borowitz [Thu, 4 Jun 2015 18:35:24 +0000 (11:35 -0700)]
Config: Distinguish between empty and null strings

The C git API and command line tools distinguish between a key having
the empty string as a value and no key being present in the config
file:

$ echo -e '[a]\nx =' > foo.config
$ git config -f foo.config a.x; echo $?

0
$ git config -f foo.config a.y; echo $?
1

Make JGit make the same distinction. This is in line with the current
Javadoc of getString, which claims to return "a String value from the
config, null if not found". It is more reasonable to interpret "x ="
in the above example as "found" rather than "missing".

We need to maintain the special handling of a key name with no "="
resolving to a boolean true, but "=" with an empty string is still not
a valid boolean.

Change-Id: If0dbb7470c524259de0b167148db87f81be2d04a

9 years agoMerge branch 'stable-4.0' 06/49406/1
Jonathan Nieder [Wed, 3 Jun 2015 19:26:54 +0000 (12:26 -0700)]
Merge branch 'stable-4.0'

* stable-4.0:
  Revert "Fix unchecked conversion warning in MergeFormatter"
  Delete deprecated TemporaryBuffer.LocalFile constructors
  Delete deprecated FileUtils.setExecute(File, boolean)
  Delete deprecated WorkingTreeIterator.isModified(DirCacheEntry, boolean)
  Delete deprecated UploadPackMayNotContinueException
  Delete deprecated TransferConfig.isFsckObjects()
  Delete deprecated TextBuiltin.out
  Delete deprecated Merger.getBaseCommit()

Change-Id: Id23a39cc6a3cd122ff1738cb85b7451bbfd8af5f
Signed-off-by: Jonathan Nieder <jrn@google.com>
9 years agoRevert "Fix unchecked conversion warning in MergeFormatter" 00/49400/1
Shawn Pearce [Wed, 3 Jun 2015 18:31:31 +0000 (14:31 -0400)]
Revert "Fix unchecked conversion warning in MergeFormatter"

This reverts commit 6d4ebd168da236c5753f237e77b2dfcbb4e42065.

Change-Id: I01f0c7cd997b9b3577cc1e98ab1adb5cdc5b55e4

9 years agoDelete deprecated TemporaryBuffer.LocalFile constructors 39/49339/1
Matthias Sohn [Sat, 25 Apr 2015 22:49:24 +0000 (00:49 +0200)]
Delete deprecated TemporaryBuffer.LocalFile constructors

Change-Id: I5fdbf997ebcbe48e86f4c41c6b2643f47054b0c3
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoDelete deprecated FileUtils.setExecute(File, boolean) 38/49338/1
Matthias Sohn [Fri, 24 Apr 2015 23:08:42 +0000 (01:08 +0200)]
Delete deprecated FileUtils.setExecute(File, boolean)

Change-Id: Ifa63e3743fb4de25e074404c301f860a1ffb5e02
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoDelete deprecated WorkingTreeIterator.isModified(DirCacheEntry, boolean) 37/49337/1
Matthias Sohn [Fri, 24 Apr 2015 23:07:20 +0000 (01:07 +0200)]
Delete deprecated WorkingTreeIterator.isModified(DirCacheEntry, boolean)

Change-Id: I687c392e5a625fd66c45998c94373aa59921b986
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>
9 years agoDelete deprecated UploadPackMayNotContinueException 36/49336/1
Matthias Sohn [Fri, 24 Apr 2015 23:05:40 +0000 (01:05 +0200)]
Delete deprecated UploadPackMayNotContinueException

Change-Id: Id298fc860684d2a2288769282fd8ce4475009dc8
Signed-off-by: Matthias Sohn <matthias.sohn@sap.com>