Author: Martin von Gagern
Revision Date: 2014-04-30 07:38:52 UTC

Document preferred-users setting in users.conf.

Author: Martin von Gagern
Revision Date: 2013-07-16 13:23:31 UTC

Switch from 12.04 (precise) to 13.04 (raring).

Firefox won't start without dbus, so we have to make sure that we have a
dbus session at that time. We also have to generate a machine-id, since
stateless systems should generate these upon boot. On the winning side, we
no longer need our own xorg server package, since touch handling in stock
server seems to work out of the box now.

Author: Martin von Gagern
Revision Date: 2013-07-15 11:02:32 UTC

Include mkfs.vfat so one can use that when installing to a new medium.

Author: Martin von Gagern
Revision Date: 2012-07-09 18:52:45 UTC

Merge from lp:ubuntu/dkms; fix conflict in changelog.

After noticing the merge conflicts in the merge request, I realized that
said request was against the main ubuntu tree, not the precise tree, so I
retargeted the changelog entry for quantal as well.

Author: Martin von Gagern
Revision Date: 2012-06-18 19:11:42 UTC

Restore compatibility for tarballs not generated by mktarball.

Author: Martin von Gagern
Revision Date: 2012-05-23 10:01:03 UTC

Beta release 1.2.0.

Author: Antonio Borneo
Revision Date: 2012-04-21 15:56:11 UTC

Fix compatibility with Solaris 10 /bin/sh

It doesn't support $( ) or $(( )) or [ -e ] or if !

Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>

Author: Martin von Gagern
Revision Date: 2012-03-01 09:19:32 UTC

merge from trunk via persist

Author: Martin von Gagern
Revision Date: 2012-03-01 09:18:21 UTC

merge from trunk

Author: Martin von Gagern
Revision Date: 2011-10-17 12:20:41 UTC

Add blackbox test for full and root-dir log in current working directory.

Author: Martin von Gagern
Revision Date: 2011-09-10 13:35:51 UTC

Test case exposing bug #842695.

When restricting a log to a given directory, _generate_deltas will be sued
to find out matching revisions. It does so using
repository.get_deltas_for_revisions, which describes the difference that the
given revision introduced with respect to its left hand parent. So files
introduced by the right hand parent of a commit will be considered "added"
by the delta. This can lead to false positives, to commits reportedly
touching a given dir although they only merged stuff introducing these
files.

Note that for in some cases, this is expected behaviour: every merge on the
route from a modification to its first merge into mainline should be
considered touched by that modification. But for stuff already included in
main line, those modifications should not be reported again if they are
merged into some side line. In other words, every change should have one
direct child reporting it, but no more.

_generate_deltas apparently processes revisions in batches of 200. After
each batch, the found additions are removed from the fileid_set and won't be
tracked in the next batch. Processing terminates if there are no more files
to track. Due to this logic, a false positive in one batch can lead to
false negatives later on, as the file gets removed too early, and its actual
addition is therefore lost.

Author: Martin von Gagern
Revision Date: 2011-09-09 16:54:14 UTC

Adjusted log catcher in blackbox test to exclude sidelines by default.

This more closely matches the behaviour of the command line tool, as is to
be expected for a blackbox test. Adjusted the TestLogMultiple class as at
least one test case in that class relies on the old behaviour.

Also found one failing test in MainlineGhostTests, which appears to be a
genuine bug. Replicated that test case to show its dependence on the levels
parameter.

Author: Martin von Gagern
Revision Date: 2011-09-07 12:15:11 UTC

Catch one more detail in #842575 test case.

I found out that recreating the file will list it as added instead of
missing, despite the ignore. So I've rearranged the tests so that the
ignore and the --take-other (which preserves the base name) are now in the
same test, and ensure that adding the file again will cause no output at
all.

Author: Martin von Gagern
Revision Date: 2011-08-29 19:14:31 UTC

Test lazy import race conditions using different stop locations.

Author: Martin von Gagern
Revision Date: 2011-08-29 16:10:23 UTC

Disallow lazy imports of anything but whole modules.

Author: Martin von Gagern
Revision Date: 2011-08-07 15:09:35 UTC

[checkbump] Adjust to new relative URI format at Google code.

Author: Martin von Gagern
Revision Date: 2011-02-08 08:54:06 UTC

Blackbox test reproducing bug #715058.

Author: Martin von Gagern
Revision Date: 2010-12-21 17:19:16 UTC

[dev-python/pykerberos]: Add support for app-crypt/heimdal

Author: Martin von Gagern
Revision Date: 2010-11-11 10:55:25 UTC

merge lp:~vila/bzr/638451-malformed

Author: Martin von Gagern
Revision Date: 2010-10-15 12:31:07 UTC

First step towards "bzr commit --amend".

Implement the --amend option and some basic support for it. The order is
uncommit - edit - commit, as the message is built on the information
collected during the commit, so it is difficult to show the correct message
for editing unless the commit is actually under way. This means that we
hold a lock while editing, and also that the message editing might fail for
various reasons after we already did the uncommit. We make some effort to
recover from the latter by restoring the uncommitted revision, but we might
end up with conflicts that way. Plenty of room for improvements.

Author: Martin von Gagern
Revision Date: 2010-06-28 20:35:52 UTC

Merge branch from bug 149270.

That branch has already been approved, and the blackbox tests from both
branches would conflict. Now they can both be merged into bzr.dev.

Author: Martin von Gagern
Revision Date: 2010-05-29 09:21:34 UTC

Fix test_revspec_tag_endrange on Mac OS X 10.5.8.

Author: Martin von Gagern
Revision Date: 2010-04-30 08:27:50 UTC

Allow different formats for tag completion.

The tests now use the dirstate-tags format, as does the
bzrlib.tests.test_tag module. As "bzr tags" will print a warning about a
deprecated format for this, we should discard its error output.

That seems a good idea for production environments as well, as outdated
formats might be quite frequent, and having completion print error messages
for all of these is certainly bad style.

Author: Martin von Gagern
Revision Date: 2010-04-23 06:43:53 UTC

Fix broken newline unescaping.

In order to show useful diffs, and readable output in general, newlines
should not be escaped. The unescaping done before was broken if r'\n'
appeared in the input, as Andrew Bennetts reviewed. In order to avoid using
regular expressions, we now split lines and join them after escaping. Added
a test case as well, and a new test class for toplevel functions in tests.

Author: Martin von Gagern
Revision Date: 2010-04-12 10:36:42 UTC

Drop outdated shell completions for bash and zsh.

There are new README files in the corresponding directories indicating
suggested alternatives instead.

Author: Martin von Gagern
Revision Date: 2010-03-15 13:19:25 UTC

Ongoing development after 1.0.2 release

Author: Martin von Gagern
Revision Date: 2009-12-09 12:48:16 UTC

Experimental revno calculation for BzrDirNode based on lefthand history.

This experimental modification reproduces pretty much the algorithm from
BzrVersionedNode.get_history: iterate through lefthand history, looking for
changesets (or in this case the latest changeset) touching a given
directory. This gets rid of get_ancestry in the get_content_revision
method.

With large subtrees which have been modified recently this algorithm will
perform much better. On the other hand, with small directories that haven't
been modified in a long time, this iteration over changesets will probably
be much more expensive than the old approach.

The resulting revision numbers will be different as well. Where the old
approach yielded the (probably dotted) revno of the last child modification
according to some topologicaly sorted but otherwise arbitrary ancestry, this
new approach uses only lefthand history and will therefore always yield
mainline revision numbers if starting from a mainline revision. This
contrasts with the revision numbers for files, which are still dotted ones.

Author: Martin von Gagern
Revision Date: 2009-11-13 17:58:46 UTC

Post-1.1 version number: 1.2-SNAPSHOT

Author: Martin von Gagern
Revision Date: 2009-07-28 14:12:13 UTC

XALANJ-2458: implement erratum 23 of the XSLT recommendation

Fixes https://issues.apache.org/jira/browse/XALANJ-2458 :
Xalan-J does not implement erratum 23 of the XSLT recommendation

For xsl:number, with level="any", if there are no matching nodes, Xalan-J
generates a text node with "0". The erratum states that, if there are no
matching nodes, there will be no text node in the result tree.

See the erratum for more details:
http://www.w3.org/1999/11/REC-xslt-19991116-errata/

The test cases have to be updated accordingly.

Author: Martin von Gagern
Revision Date: 2009-07-28 17:38:57 UTC

XALANJ-2346: fixes for faulty NaN handling
Fixes https://issues.apache.org/jira/browse/XALANJ-2346
Patch by Dave Brosius

Author: dbertoni
Revision Date: 2008-08-15 23:57:23 UTC

Added test case for Jira issue XALANC-680.

Author: zongaro
Revision Date: 2008-06-19 02:40:41 UTC

Part of fix for Jira issue XALANJ-2446:

The implementation of Parser.getQName(String,String,String) method was ensuring
that any particular pair of namespace URI and local part of the name would
result in a unique QName object. It did not take into account the prefix,
despite the fact that the original prefix might be associated with a different
namespace URI in this new context or no namespace at all.

Fixed this by making getQName take into account all three of the namespace URI,
local part of the name and prefix in order to create unique QName objects.

Some other parts of XSLTC depended on the assumption that a QName object
uniquely identified a pair consisting of the namespace URI and local part of
the name and used object reference comparisons (== or !=) on QName objects.
These comparisons all had to change to use equals(Object) for comparison
instead. The implementation of QName.equals(Object) itself had to change to
consider the namespace URI and local part of the name rather than relying on
object identity.

Part of fix for Jira issue XALANJ-2447:

The handling of exclude-result-prefixes for an xsl:stylesheet had the complete
set of excluded prefixes accumulated on the xsl:stylesheet for the main
stylesheet module. However, the scope of the attribute is supposed to consist
only of elements that are direct descendants of an element that has the
attribute, and does not extend to imported or included stylesheets. The same
is true of namespaces excluded because they have been declared as extension
element prefixes.

Fixed this by maintaining a stack of excluded prefixes that's pushed on entry
to the Stylesheet.parseOwnChildren method and popped on exit from the same.
The push operation clears the current set of excluded prefixes and the pop
operation restores the set of excluded prefixes that were in effect for the
importing or including stylesheet.

Reviewed by Christine Li (jycli () ca ! ibm ! com)

Author: Martin von Gagern
Revision Date: 2008-12-18 11:17:47 UTC

merged from trunk via persist

Author: Martin von Gagern
Revision Date: 2008-06-17 09:03:05 UTC

Skip literals to keep the noise down. The approach is rather heuristic, and
might yield both false positives (lines on the stack contain the string as a
literal, but the converted object comes from a variable instead) and false
negatives (e.g. different use of escapes). Most occurrences are handled
correctly, though, which increases the speeed and usability.

Author: Martin von Gagern
Revision Date: 2008-06-07 10:05:28 UTC

Rename variables (s/sha_bang/lircrcd_cmd/) to beter reflect usage.