hi, patch attached :)
it adds two new testcases: * new test: test-darcs2git-move-recursive-and-modify.sh given the following patch: rename dir1 -> dir2; modify dir2/file this testcase ensures tailor modifies dir2/file, and not dir1/file. and * new test: test-darcs2git-accent.sh this is a testcase for the bug which was fixed by the patch "darcs source: handle unescaped problematic characters"
New patches:
[new test: test-darcs2git-move-recursive-and-modify.sh
Miklos Vajna <[EMAIL PROTECTED]>**20080405213033
given the following patch:
rename dir1 -> dir2; modify dir2/file
this testcase ensures tailor modifies dir2/file, and not dir1/file.
] {
addfile ./test-scripts/test-darcs2git-move-recursive-and-modify.sh
hunk ./test-scripts/test-darcs2git-move-recursive-and-modify.sh 1
+#!/bin/sh
+
+# File: test-darcs2git-move-recursive-and-modify.sh
+# needs: test-darcs2git.include
+#
+# Test for converting from Darcs to Git.
+# Tests if we can update a file which is just moved recursively within
+# this patch.
+
+. ./test-darcs2git.include
+darcs_setup
+
+mkdir -p dir1/subdir
+echo foo >dir1/subdir/file
+darcs add dir1/subdir/file
+darcs record -v -a -A Nobody -m "import file"
+
+mkdir dir2
+darcs add dir2
+darcs mv dir1/subdir dir2
+echo bar >> dir2/subdir/file
+darcs record -v -a -A Nobody -m "move subdir to dir2 and update file"
+
+testing_runs
}
[new test: test-darcs2git-accent.sh
Miklos Vajna <[EMAIL PROTECTED]>**20080405213329
this is a testcase for the bug which was fixed by the patch "darcs
source: handle unescaped problematic characters"
] {
addfile ./test-scripts/test-darcs2git-accent.sh
hunk ./test-scripts/test-darcs2git-accent.sh 1
+#!/bin/sh
+
+# File: test-darcs2git-accent.sh
+# needs: test-darcs2git.include
+#
+# Test for converting from Darcs to Git.
+# This test ensures tailor won't fail if there are accents in the author
+# name.
+
+. ./test-darcs2git.include
+darcs_setup
+
+echo "foo" > foo
+darcs add foo
+darcs record -v -a -A "Nobodyéáőű" -m "add foo"
+
+testing_runs
}
Context:
[Added a note about patch-name-format and its implication in potential hash
collisions under darcs
[EMAIL PROTECTED]
[Factorize the sources and targets in the example
[EMAIL PROTECTED]
Given that all projects migrate patches from Monotone to Subversion,
take advantage of the "source = kind:" syntax.
]
[reflect test capability in usage hint
[EMAIL PROTECTED]
[Fix exception handling typo
Alexey Morozov <[EMAIL PROTECTED]>**20080327105552]
[darcs source: don't fail on empty author
Miklos Vajna <[EMAIL PROTECTED]>**20080324145121
unfortunately darcs allows recording a patch with no author info, so we
should not fail in that case.
]
[new test: ./test-darcs2git-add-modify.sh
Miklos Vajna <[EMAIL PROTECTED]>**20080322021527
this is a very basic test: just adds a new file and then updates it. i
think it's important to have such a simple test, not just complex ones,
so that if this fails as well, one can debug just this and not a complex
one.
]
[add first darcs2git test
Miklos Vajna <[EMAIL PROTECTED]>**20080322020913
this tests the bug fixed in "DarcsChangeset.addEntry: fix optimization"
]
[git target: avoid git mv in general cases
Miklos Vajna <[EMAIL PROTECTED]>**20080321034550
if a direcotory is removed and files are renamed + modified under the
given dir, we can't use git mv as it will look for the old path of the
files. but just using git add -u before the command and using git add
for the new files works perfectly.
]
[git target: handle renaming of empty directories
Miklos Vajna <[EMAIL PROTECTED]>**20080321021553]
[git target: don't be too agressive when detecing the a -> a/b case
Miklos Vajna <[EMAIL PROTECTED]>**20080321020604]
[small fix in SynchronizableTargetWorkingDir._renameEntries()
Miklos Vajna <[EMAIL PROTECTED]>**20080321015625
handle the case when the directory is moved to itself. in this case the
target already removes the -TAILOR-HACKED-TEMP-NAME file.
]
[git target: rename() won't work for rename(a/b, a)
Miklos Vajna <[EMAIL PROTECTED]>**20080321015009
so do it manually
]
[darcs source: collapse "rename A B; rename B C" into "rename A C"
Miklos Vajna <[EMAIL PROTECTED]>**20080321004216]
[DarcsChangeset.addEntry: fix optimization
Miklos Vajna <[EMAIL PROTECTED]>**20080321002504
in case the changeset is like "mv a a2; mkdir a; mv a2 a/b" disable an
optimization that would break the conversion
]
[DarcsXMLChangesHandler.startElement: fix a typo
Miklos Vajna <[EMAIL PROTECTED]>**20080320225309
bool(attributes['inverted']) will give true for "True" and "False" as
well, so that's incorrect.
]
[git target: fix typo in a comment
Miklos Vajna <[EMAIL PROTECTED]>**20080315222755
i obviously wanted to write 'git mv', not 'svn mv'
]
[darcs source: improve fix for rollbacked moves + file updates
Miklos Vajna <[EMAIL PROTECTED]>**20080308190959
here is the scenario:
there is a rename: /foo -> /bar and an independent file:
/dir/foo/file.c. in this case we should not alter the file's path.
]
[darcs source: handle inverted renames + modifications
Miklos Vajna <[EMAIL PROTECTED]>**20080307230104
here is the scenario:
UNDO: move to extra + upd
./source/xapps-extra/mtools -> ./source/xapps/mtools
M ./source/xapps-extra/mtools/FB -1
* move to extra + upd
./source/xapps/mtools -> ./source/xapps-extra/mtools
M ./source/xapps-extra/mtools/FB +1
now in xml this will look like:
<patch inverted='True'
hash='20070625100419-e2957-5f39752481ac1f758883ab8a17b03a3a47606cbb.gz'>
<name>move to extra + upd</name>
<summary>
<move from="source/xapps-extra/mtools" to="source/xapps/mtools"/>
<modify_file>
source/xapps-extra/mtools/FB<removed_lines num='1'/>
</modify_file>
</summary>
</patch>
<patch inverted='False'
hash='20070625100419-e2957-16657621d68a3b47d3ad0f0153ce336f63ec2d5d.gz'>
<name>move to extra + upd</name>
<summary>
<move from="source/xapps/mtools" to="source/xapps-extra/mtools"/>
<modify_file>
source/xapps-extra/mtools/FB<added_lines num='1'/>
</modify_file>
</summary>
</patch>
so in short the xml file contains the rename list at the start of the
entry list, while the modify_file tags still refer to the old path
names. handle this case by updating the paths in the modify_file tags.
]
[git target: use git-mv
Miklos Vajna <[EMAIL PROTECTED]>**20080307225513
this is just a cleanup: git has an mv command for a long time that does
all the tricks to handle renames, we just have to do the same trick as
with svn.
]
[M-x untabify
[EMAIL PROTECTED]
[Runs with Python 2.5 too
[EMAIL PROTECTED]
[Removed ascii-art on two-way-sync
[EMAIL PROTECTED]
[git target: handle author and date information for tags properly
Miklos Vajna <[EMAIL PROTECTED]>**20080302214138
before:
$ git show 1_0|head -n3
tag 1_0
Tagger: VMiklos <[EMAIL PROTECTED](none)>
Date: Sun Mar 2 22:39:36 2008 +0100
after:
$ git show 1_0|head -n 3
tag 1_0
Tagger: T Agger <[EMAIL PROTECTED]>
Date: Sat Mar 1 22:27:13 2008 +0000
]
[git target: don't exit with an error if the working directory is already
created
Miklos Vajna <[EMAIL PROTECTED]>**20080101234424
creating the working dir twice then claiming about it already exists is not
something we want
]
[Fix #43: take advantage of Trac .rst handling
[EMAIL PROTECTED]
[Update Walter's email
[EMAIL PROTECTED]
[Whitespace
[EMAIL PROTECTED]
[Assign basedir earlier, git backend needs it at _load() time
[EMAIL PROTECTED]
[Avoid using multilevel -d option on CVS checkout
[EMAIL PROTECTED]
This fixes a problem I have with using tailor to move changesets from
one cvs repo to another.
It also sounds like ticket #101 on the trac pages.
]
[Perform a clobbering clean merge instead of a simple update under hg 0.9.5
[EMAIL PROTECTED]
[Use only the first five components from changelog.read()
[EMAIL PROTECTED]
[Replacement for localrepository.update()
[EMAIL PROTECTED]
In 0.9.5+ localrepository.update(self, node) disappeared. Quick and
dirty as it is, this patch reimplant the method using hg.update().
]
[TAG Version 0.9.30
[EMAIL PROTECTED]
Patch bundle hash:
6a08dbbaac2f71f93efe139bdcac0079999d40fe
pgpJx1H2Xz85v.pgp
Description: PGP signature
_______________________________________________ Tailor mailing list [email protected] http://lists.zooko.com/mailman/listinfo/tailor
