Skip to main content
AgeCommit message (Collapse)AuthorFilesLines
2010-09-15Qualify post-0.9.3 buildsstable-0.9Matthias Sohn2-11/+11
Change-Id: I2d496343dbcf6d74cd2214fe14e0cfe1fa346ada Signed-off-by: Matthias Sohn <>
2010-09-15EGit 0.9.3v0.9.3Matthias Sohn2-11/+11
Change-Id: I08460e3688d074278cad826b7193b622b8581392 Signed-off-by: Matthias Sohn <>
2010-09-15Qualify post-0.9.1 buildsMatthias Sohn2-11/+11
Change-Id: I28529d8df1d2992607ff229db5b6045015790018 Signed-off-by: Matthias Sohn <>
2010-09-15EGit 0.9.1v0.9.1Matthias Sohn2-11/+11
Change-Id: I46b5f2c26467365aafc0300499e8e404ebb25b8f Signed-off-by: Matthias Sohn <>
2010-09-09Sort roots by full path to ensure stable assertion resultsMatthias Sohn1-3/+13
GitResourceVariantTree uses a HashSet to store resource variants which doesn't guarantee stable sort order which may break the assertions in GitResourceVariantTreeTest.shouldReturnTwoRoots() [1]. To fix this we sort the roots before doing assertions. [1] Change-Id: If92b0bb43bfb0a022df6fefb98b06b00ace03c35 Signed-off-by: Matthias Sohn <>
2010-09-08Clear RepositoryCache after each test to fix core testsJens Baumgart1-0/+4
With this fix the core test suite is running also on Mac. Reason for the failures before was the usage of out-dated Repository instances contained in RepositoryCache which caused spurious failures when running the full test suite while running tests individually always succeeded. A follow-up fix should care for the life cycle of Repository objects in RepositoryCache. Change-Id: I1f16a9de48435bd98e2520e34e49bc1ca6893f14 Signed-off-by: Jens Baumgart <> Signed-off-by: Matthias Sohn <>
2010-09-07Merge "Fix deletion problem in EGit Core tests"Matthias Sohn1-4/+21
2010-09-06Fix deletion problem in EGit Core testsJens Baumgart1-4/+21
EGit core tests sometimes fail because TestUtils.deleteRecursive can not delete a file. To workaround the problem the deletion is retried 10 times. Change-Id: Ia25c9304884c95f54e8e7b98ed49dd044fa7b95b Signed-off-by: Jens Baumgart <>
2010-09-06Run core tests head-less also from MavenMatthias Sohn1-1/+2
From Eclipse we run the egit core tests head-less on plain osgi so there is no need to run them in UI thread nor using UI harness when running these tests from Maven. Change-Id: I84d86b66b2da3a5a1af00b14e1b12ea27ed11aaa Signed-off-by: Matthias Sohn <>
2010-09-06Merge "Fix AddOperationTest"Matthias Sohn1-1/+1
2010-09-06Merge "Fix IgnoreOperationTest"Matthias Sohn1-0/+10
2010-09-06Fix AddOperationTestJens Baumgart1-1/+1
Fixed small bug in AddOperationTest. Change-Id: Ib15b43cf9341f669911fcf65671e139373eb13ee Signed-off-by: Jens Baumgart <>
2010-09-06Fix IgnoreOperationTestJens Baumgart1-0/+10
IgnoreOperationTest did not clean up .gitignore in the workspace folder. This caused subsequent test to fail. Change-Id: Icc3e3e02302a8189ae2009e24682206d8bff086e Signed-off-by: Jens Baumgart <>
2010-09-04cleanup: Remove unnecessary @SuppressWarningsRobin Rosenberg1-8/+0
Change-Id: Ic9a424cfdfcebbf2ff548f6741d2f96a1494df0e Signed-off-by: Robin Rosenberg <>
2010-09-01Fix EGit to use JGit's api.errors packageShawn O. Pearce2-6/+7
Change-Id: I27ec972b0d42b495c1f0c26ec0d76e644224e590 Signed-off-by: Shawn O. Pearce <>
2010-08-31Remove usage of deprecated PersonIdent constructorMathias Kinzler1-2/+2
Change-Id: I7706a916f8388bbae948d550005584cadb1e836a Signed-off-by: Mathias Kinzler <>
2010-08-30Add tests for IgnoreOperationBenjamin Muskalla3-6/+130
In order to have a better test coverage before fixing and refactoring other changes, we need to ensure to have at least most of the features covered by tests. In addition, fixed minor bug to really use the scheduling rules we create. Change-Id: If63a9632b28aeb7d6ef585f889876ec42dac2c0c Signed-off-by: Benjamin Muskalla <> Signed-off-by: Chris Aniszczyk <>
2010-08-28Revert "Hide Maven target directories from Eclipse"Robin Rosenberg1-11/+0
This reverts commit 9425360ad8b57aa7020be44481ad6ed771a7261c since it breaks compatibility with Eclipse 3.5 which can no longer import the projects Bug: 323390 Change-Id: Icda21a1e00a63b1e7333b89525af04d6a776b178 Signed-off-by: Robin Rosenberg <>
2010-08-25Cleanup of HistoryTest and IndexFileRevisionBenjamin Muskalla1-73/+47
Fixed several warnings, added testcase for the IndexFileRevision and rewrote test setup to use porcelain API instead of the plumbing stuff. Change-Id: Ia568de2231e12f62f8415ca32c8aba5c89abf7b5 Signed-off-by: Benjamin Muskalla <> Signed-off-by: Chris Aniszczyk <>
2010-08-23Remove ObjectWriter and replace it with ObjectInserterShawn O. Pearce2-25/+49
JGit won't support the deprecated ObjectWriter API anymore, so we have to use the newer ObjectInserter style throughout EGit. Change-Id: I95e2c705af5d7ff6c2e7da78ec3a60a1c718708e Signed-off-by: Shawn O. Pearce <>
2010-08-23Adapt to JGit's CommitBuilder, TagBuilder API changeShawn O. Pearce3-7/+7
Change-Id: I5d69fca2b8a168b68df7d93b23f5705c49638011 Signed-off-by: Shawn O. Pearce <>
2010-08-20Replace deprecated Tag, Commit usage with RevTag, RevCommitShawn O. Pearce3-17/+16
JGit is changing its API to no longer make Tag and Commit classes accessible for reading purposes. The only thing they will support is the creation of new objects, and even then their APIs are moving to a simpler and more consistent usage style. Replace relevant uses of tags and commits with their RevWalk based variants, fixing any build breakages caused by this JGit change. Change-Id: I070e309894989024b26be7638bbfcc62a9a4b00d Signed-off-by: Shawn O. Pearce <>
2010-08-18Keep up with Tag API changesChris Aniszczyk1-1/+9
The Tag API from JGit was changed to remove duplication. Change-Id: I562198f7a6c21b659e78339ef8c4d9d2bff248ab Signed-off-by: Chris Aniszczyk <>
2010-08-12Fix for bug in GitResourceVariantComparatorDariusz Luksza1-2/+22
When comparing local folder with folder that wasn't exist remotly GitResourceVaraintCompare return that both resources are equal. I've added additional condition for resource existence. Change-Id: Ie8b97fffd1237723afce6aed3c116b280694b3ba Signed-off-by: Dariusz Luksza <>
2010-08-11Improve GitResourceVariant performanceDariusz Luksza1-23/+29
Recursive TreeWalk's for obtaining objectId were replaced by additional onstructor parameter, this change should improve performance of synchronize action. Change-Id: I6c221ee7a1f754262d008bffe5103d37082773dc Signed-off-by: Dariusz Luksza <> Signed-off-by: Chris Aniszczyk <>
2010-08-11Update of GitSynchronizeData APIDariusz Luksza2-21/+37
In Git synchronization we always operate on commit level changes therefore source and destination points should be always RevCommit objects. In case of synchronizing tags we should map given tag with proper RevCommit object. Instead of doing this multiple times on different levels I've add this resolution to GitSynchronizeData constructor. Change-Id: I6476b4940783d4b719d08d9c0c5bb3060b1ee412 Signed-off-by: Dariusz Luksza <> Signed-off-by: Chris Aniszczyk <>
2010-08-11Merge "Preliminary support for ChangeSets"Chris Aniszczyk5-12/+20
2010-08-11Merge "Hide Maven target directories from Eclipse"Chris Aniszczyk1-0/+11
2010-08-10Preliminary support for ChangeSetsDariusz Luksza5-12/+20
Adds proper representation of git changes that occurs between two ref's. Changes are displayed as Team ChangeSet instance. All commits are listed as a child's of node that represent repository. Every commit node contains tree-list of all changed, added or deleted resources. Bug: 318473 CQ: 4400 Change-Id: I198bb3859463ee7e0e230e22cf06761391436cd8 Signed-off-by: Chris Aniszczyk <> Signed-off-by: Dariusz Luksza <>
2010-08-10Merge "Fixed several warnings around ConnectProviderOperation"Mathias Kinzler2-29/+11
2010-08-09Hide Maven target directories from EclipseRobin Rosenberg1-0/+11
Change-Id: I804c8b4b4bcc907226dac1ea414a9eabad8bdc1c Signed-off-by: Robin Rosenberg <>
2010-08-06Set correct date when committing a mergeStefan Lay2-0/+121
The date was not set before. This is corrected with this commit. A test case was added. Change-Id: Icd4f750e84a4fb74133b05fdeec57a7faf612547 Signed-off-by: Stefan Lay <>
2010-08-05Fixes obtaining common ancestor in Synchronize viewDariusz Luksza9-1098/+980
For obtaining common base ancestor I've used suggested RevWalk with RevFilter.MERGE_BASE. For API compatibility it is also used for obtaining remote variant. This patch also improves the GitResourceVariant implementation. I replaced all dependencies to the deprecated TreeEntry with the proper RevObject implementation. Bug: 317934 CQ: 4390 Change-Id: I8f5352862d01e5a5475a74f222241970cd7960ee Signed-off-by: Dariusz Luksza <> Signed-off-by: Chris Aniszczyk <>
2010-08-05Fixed several warnings around ConnectProviderOperationBenjamin Muskalla2-29/+11
Removed suspect usage of Job to wait for a test result even though the operation in question can run synchronously. Removed parameter assignment in Operation and replaced ObjectWriter with ObjectInserter. Change-Id: I578d2e64d1fe5181395c518329b339339051d870 Signed-off-by: Benjamin Muskalla <>
2010-08-05Cleanup temp dirs after test executionBenjamin Muskalla6-17/+18
Currently the EGit Core Suite fails to clean up the temp directories in my user dir. Instead of only deleting the working dirs, we now use the deleteTempDirs to ensure we cleanup everything that we create during the testrun. Also renamed getTempDir to createTempDir to reflect the real intention. Change-Id: I949e36ba0ae18b537fc9d5424af13fa8f2613046 Signed-off-by: Benjamin Muskalla <>
2010-07-30Fixed deprecated API usageBenjamin Muskalla1-1/+2
Besides fixing the deprecated API call, also moved the sleep in AppOperationTest to a more reasonable location. For me, the tests failed most of the time as localTime and the timestamp in the repo were the same. I think it makes more sense to wait after the commit to have a time gap between commit and file change Change-Id: I3c481096f1b89f35b0b232fbcd8f4b1ad0d012fc Signed-off-by: Benjamin Muskalla <>
2010-07-30Run egit.core suite headlessBenjamin Muskalla1-1/+3
Without a product and the right application, the egit.core suite can run headless which reduces the time running all the tests. As core has to dependencies to the UI, we can easily ignore all the workbench startup and just run on plain OSGi. Change-Id: I8e200c7aef0e36481e31ee04dcf3440c1d12f2b8 Signed-off-by: Benjamin Muskalla <>
2010-07-29AddOperationTest uses 3.6 APIMathias Kinzler1-2/+2
We still must run on 3.5, so we can't use 3.6 specific calls. Change-Id: Ie6d40b2a57bb9cd626d8dcb42f0a11662f47947e Signed-off-by: Mathias Kinzler <>
2010-07-29Remove menu items "Track" and "Refresh"Stefan Lay3-12/+8
Now the add command also adds new files to the repository. Additionally it takes into account .gitignore. Therefore there is no need anymore of the "Track" and "Refresh" menu items. Bug: 315306 Change-Id: I85ec194d0bb89a895a3ff459061ed6f3152b1943 Signed-off-by: Stefan Lay <>
2010-07-28Make use of JGit Add porcelain commandStefan Lay4-6/+225
Now whole subtrees can be added by the Add menu item. It stages changes, adds files and resolves conflicts. Files ignored by .gitignore or by Eclipse are not added to the repository. Change-Id: I23ddb4be38d668f675f2fc936863dc4e72822f14 Signed-off-by: Stefan Lay <>
2010-07-27Update to latest JGit APIShawn O. Pearce12-26/+34
* Change to use FileRepository throughout most of EGit. Too much of the code currently has assumptions about the repository being on the local filesystem, accessible by Use the file specific repository type to better describe the dependency. * Replace getWorkDir() with getWorkTree() * Replace openObject(), openBlob(), hasObject() with the newer short style open() and has(). * Always treat blobs as InputStreams when possible. This ensures that even on very large files, EGit won't explode the JVM heap and cause an OutOfMemoryError. * Adapted to the new style RepositoryListener interfaces, which promises better long-term API stability. * Replace static DirCache read() and lock() with Repository instance methods, further hiding the storage method of the working directory's index. Change-Id: I2459757a26f8c724b95468f9f38bac2ee172a932 Signed-off-by: Shawn O. Pearce <>
2010-07-16Fix PushOperationTestMathias Kinzler1-0/+6
This currently fails on file systems with low time stamp accuracy or a Java implemenration that does not return the subseconds part of the filestamp. Java on Linux exhibits this bug. See Change-Id: If54da3e1209ca2ca6e11283b9c08d7b7504f7845 Signed-off-by: Mathias Kinzler <> Signed-off-by: Robin Rosenberg <>
2010-07-09Remove some deprecated JGit API from CoreTestsMathias Kinzler3-17/+18
Bug: 319152 Change-Id: Icb91f6f8fe6d385d4d6577f1cfe3568f32595cd1 Signed-off-by: Mathias Kinzler <>
2010-07-09Merge "EGit Core Tests: cleanup"Stefan Lay5-8/+5
2010-07-09Stabilize Core TestsBenjamin Muskalla3-18/+48
There are currently failures when the test workspace is not cleared. In the TrackUntrackOperation test, apparently, the check whether a file is "tracked" was too simplistic. Other fixes were required because resources were not handled properly. With these fixes, it is possible to run the tests without clearing the test workspace. The fixes have successfully been tested on Helios, too. Bug: 318444 Change-Id: I31cbbb63b5aa345d3f4afdea1078640973bbbbac Signed-off-by: Mathias Kinzler <>
2010-07-07EGit Core Tests: cleanupMathias Kinzler5-8/+5
1. Get rid of Txxxx-prefixes for test classes 2. Remove System.out.println() calls which don't help but disturb the test output (in ConnectProviderOprationTest.testNewUnsharedFile()) No tests logic was changed. Change-Id: Ie11249e0a3be19ea6a09319d4daae4134a6342f4 Signed-off-by: Mathias Kinzler <>
2010-07-07EGit Core tests: fix PushOperation and add testsMathias Kinzler13-24/+1000
While writing tests for EGit Core, the following problems with PushOperation were identified: 1. an operation result is returned even if the operation was not run before 2. when re-using he specification used to create the operation, the operation result is completely wrong (says UP_TO_DATE and does not update the remote repo even if it should) These issues were fixed and corresponding tests were added. Other tests for the 1. ListRemoteOperation 2. TagOperation 3. TrackOperation 4. UntrackOperation 5. DiscardChangesOperation were added along with some infrastructure for these tests (which need to set up two repositories instead of just one). There were also some intermittent failures to delete resources in GitResourceVariant* and BlotStorage tests. Hopfully the change there will improve this. Total test coverage for egit core improves from around 34 percent to around 51 percent. Bug: 318444 Change-Id: I80a79260666724ef3ebd5de097790a433049fee0 Signed-off-by: Mathias Kinzler <>
2010-06-30fixed usage of Java 1.6 APIBenjamin Muskalla1-1/+2
Bug: 318525 Change-Id: I0989012db58125832f207e33ab6bf5030541c276
2010-06-30Add tests for integration with Synchronize viewDariusz Luksza6-3/+1821
Adds test cases for GitSynchInfo.getKind(), GitResourceVariantComparator .compare() methods and SWTBot tests for UI testing of Synchronize View. Bug: 316957 CQ: 4275 CQ: 4291 Change-Id: I169d38c33102494ac7f9dfc5e274aec1392850d9 Signed-off-by: Dariusz Luksza <> Signed-off-by: Matthias Sohn <>
2010-06-21Add Test for ResetOperationJens Baumgart3-63/+376
A test for ResetOperation was added. Common test functionality for creating a test repository was moved to a new class TestRepository to enable reuse in other tests. Also a missing license header was added. Change-Id: I5f6852cb8d85b434720d052e05b7c8f8157f88a6 Signed-off-by: Jens Baumgart <> Signed-off-by: Matthias Sohn <>

Back to the top