Alexander Kitaev <Alexander.Kitaev@svnkit.com>


Found Patch Suggested
r879747, r872068, r871388, r871385, r858768 r920363, r920351, r851376, r851371 r920343

r920363 | markphip | 2010-03-08 09:47:57 -0600 (Mon, 08 Mar 2010)

JavaHL: Change constructors from package protected to public so that
alternate JavaHL implementations like SVNKit do not have to use our package
namespace just so that they can construct these classes.

Patch by: Alexander Kitaev

[ in subversion/bindings/javahl/ ]
* src/org/apache/subversion/javahl/ChangePath.java:
* src/org/apache/subversion/javahl/CommitItem.java:
* src/org/apache/subversion/javahl/ConflictDescriptor.java:
* src/org/apache/subversion/javahl/ConflictVersion.java:
* src/org/apache/subversion/javahl/DiffSummary.java:
* src/org/apache/subversion/javahl/DirEntry.java:
* src/org/apache/subversion/javahl/Info.java:
* src/org/apache/subversion/javahl/Info2.java:
* src/org/apache/subversion/javahl/Lock.java:
* src/org/apache/subversion/javahl/LogMessage.java:
* src/org/apache/subversion/javahl/NotifyInformation.java:
* src/org/apache/subversion/javahl/ProgressEvent.java:
  Make constructor public.

r920351 | markphip | 2010-03-08 09:14:01 -0600 (Mon, 08 Mar 2010)

JavaHL: Change NativeResources.version to private and add a public
getVersion()  method to retrieve it.  This was suggested by SVNKit to
facilitate supporting our new and old namespace in their implementation.

Patch by: Alexander Kitaev

[ in subversion/bindings/javahl/ ]
* src/org/apache/subversion/javahl/NativeResources.java:
  Make version private and add public getVersion method.

* src/org/apache/subversion/javahl/SVNAdmin.java:
* src/org/apache/subversion/javahl/SVNClient.java:
* src/org/tigris/subversion/javahl/SVNAdmin.java:
* src/org/tigris/subversion/javahl/SVNClient.java:
  Use NativeResources.getVersion() instead of accessing field
  directly.


r920343 | markphip | 2010-03-08 09:03:57 -0600 (Mon, 08 Mar 2010)

JavaHL: Extract interface from SVNAdmin class to make it easier for
alternative implementations to exist.

Suggested by: Alexander Kitaev

[ in subversion/bindings/javahl/ ]

* src/org/apache/subversion/javahl/ISVNAdmin.java
  New interface class for SVNAdmin methods.

* src/org/apache/subversion/javahl/SVNAdmin.java
  Updated to implement ISVNAdmin.

* tests/org/apache/subversion/javahl/SVNTests.java:
  Reference ISVNAdmin instead of SVNAdmin.


r879747 | stsp | 2009-09-29 10:08:47 -0500 (Tue, 29 Sep 2009)

* subversion/libsvn_wc/update_editor.c
  (add_file): Don't pass NULL pointers to strcmp. There's a known way
    to trigger this crash in 1.6.x, and in trunk it's more or less
    waiting to happen.

Found by: Alexander Kitaev <Alexander.Kitaev@svnkit.com>


r872068 | dlr | 2008-07-03 19:58:46 -0500 (Thu, 03 Jul 2008)

JavaHL: Fix bug in the CopySources::array() JNI method, where the peg
revision was extracted incorrectly from a Java object because we were
calling the wrong method (we were using the revision instead).

[ in subversion/bindings/javahl/ ]

* native/CopySources.cpp
  (array): Invoke the CopySource.getPegRevision() method when
    extracting the peg revision of a copy source from a Java object.

Reported by: Alexander Kitaev <Alexander.Kitaev@svnkit.com>


r871388 | kfogel | 2008-05-20 13:08:11 -0500 (Tue, 20 May 2008)

Fix bug whereby 'svn cp --parents URL/TO/EXISTING/FILE URL/TO/EXISTING/DIR'
would seg fault.  This follows up to the regression test added in r31311.

Found by: Alexander Kitaev <alex@tmate.org>

* subversion/libsvn_client/copy.c
  (repos_to_repos_copy): Specially handle the case where dir is an
    immediate child of top_url.

* subversion/tests/cmdline/copy_tests.py
  (test_list): Expect 'unneeded_parents' to pass now.


r871385 | kfogel | 2008-05-20 10:36:39 -0500 (Tue, 20 May 2008)

Add a regression test for a bug that will be fixed in r31314.

Found by: Alexander Kitaev <alex@tmate.org>

* subversion/tests/cmdline/copy_tests.py
  (unneeded_parents): New test.
  (test_list): Run it.


r858768 | lundblad | 2006-03-02 16:53:28 -0600 (Thu, 02 Mar 2006)

Fix a bug whereby svn unlock would unlock a path even if the wrong
lock token is stored in the WC

Found by: Alexander Kitaev <alex@tmate.org>
Suggested by: jerenkrantz
(Some initial analyzes.)

* subversion/libsvn_client/locking_commands.c
  (organize_lock_targets): Get rid of the parent_entry_p argument; callers
  updated.
  (svn_client_lock): Remove bogus check for whether the common parent
  URL really is a URL (Uh! it always is) and pass the WC path when opening
  the RA session if available (which was the intention before as well).
  (svn_client_unlock): Dito. Only fetch lock tokens from the server when the
  user gave us URLs.

* subversion/tests/cmdline/lock_tests.py
  (unlock_wrong_token): Check for specific (though weird) error messages.
  (test_list): Remove XFail for unlock_wrong_token.


r851376 | kfogel | 2004-10-10 00:13:26 -0500 (Sun, 10 Oct 2004)

* www/project_links.html: Add link to JavaSVN.  

Patch by: Alexander Kitaev <alex@tmate.org>


r851371 | ehu | 2004-10-09 03:48:40 -0500 (Sat, 09 Oct 2004)

* www/project_links.html: Add JavaSVN.  

Patch by: Alexander Kitaev <alex@tmate.org>