| Age | Commit message (Collapse) | Author |
|---|
|
|
|
being a special case of the commit method; includes tests
util: Fixed iterable lists, which broke due to an incorrectly implemented __contains__ method
|
|
the ORIG_HEAD reference as well
|
|
able to handle submodule removals, additions, path changes and branch changes. All this still needs to be tested though
|
|
offering a create_* method for most important entities.
Moved implementation of smart update method to the RootModule implementation, where it may do special things without requiring an interface for everything
|
|
property alrdeady
Improved submodule move tests
|
|
to-be-improved test
|
|
of attribute lookups considerably
|
|
|
|
coverage for it
|
|
|
|
fixed a few issues on the way
|
|
next
|
|
|
|
to update submodules such as svn-externals
|
|
|
|
heads, including test.
Config: SectionConstraint was updated with additional callable methods, the complete ConfigParser interface should be covered now
Remote: refs methods is much more efficient now as it will set the search path to the directory containing the remote refs - previously it used the remotes/ base directory and pruned the search result
|
|
though to allow easy configuration of branch-specific settings
|
|
svn-external like behaviour. Implemented first version of update, which works for now, but probably needs to see more features
|
|
general may be contradicting if a tag is given there, as well as a commit sha of the submodule. Hence it should really be only a branch
|
|
was mainly copy-paste from with_rw_repo, what a shame
|
|
provided for Remotes, including test
|
|
local cache - previously a procedural approach was used, which was less code, but slower too. Especially in case of CommitObjects unrolling the loop manually makes a difference.
Submodule: Implemented query methods and did a bit of testing. More is to come, but the test works for now. As special addition, the submodule implementation uses the section name as submodule ID even though it seems to be just the path. This allows to make renames easier
|
|
corresponding locks. Submodule class now operates on parent_commits, the configuration is either streamed from the repository or written directly into a blob ( or file ) dependending on whether we have a working tree checkout or not which matches our parent_commit
|
|
usable. It showed that the ConfigParser needs some work. If the root is set, it also needs to refer to the root_commit instead of to the root-tree, as it will have to decide whether it works on the working tree's version of the .gitmodules file or the one in the repository
|
|
Bryan Bishop
|
|
dependent on the setup of the surrounding repository, hence the amount of ref-types found is actually variable, as long as they get more
|
|
has better testing for the use of paths during reset. The IndexFile now implements this on its own, which also allows for something equivalent to git-reset --hard -- <paths>, which is not possible in the git command for some probably very good reason
|
|
instead of the existing and valid. The rest of the ConfigParser handling is correct, as it reads all configuration files available to git
see http://github.com/Byron/GitPython/issues#issue/1
|
|
head.reset: will now handle resets with paths much better, especially in the --mixed case, see http://github.com/Byron/GitPython/issues#issue/2
|
|
http://groups.google.com/group/git-python/browse_thread/thread/84ed1835e26a5296?hl=en
|
|
Fixed test which used the --force flag on move, but there is only a short version (left) it appears
|
|
previously, although there was absolutely no need for that.
See http://byronimo.lighthouseapp.com/projects/51787/tickets/41-diff-regex-lib_git_diffpy-cannot-handle-paths-with-spaces
|
|
|
|
into the byte stream, as well as decoded from it
|
|
passed in
test_odb: added more information to the message output
|
|
http://byronimo.lighthouseapp.com/projects/51787/tickets/44-remoteref-fails-when-there-is-character-in-the-name using supplied patch ( which was manually applied ).
Fixed slightly broken test for remote handling
|
|
|
|
written according to the encoding of the commit object, and decoded using that information as well. Trees will encode and decode their names with utf8
|
|
repo.clone: assured backslashes won't reach the remote configuration, as it can cause trouble when re-reading the file later on. Some git commands don't appear to be able to properly deal with backslashes, other's do
|
|
return value
remote: fixed evil bug that was caused by some inconsistency of python when __getattr__ and __slots__ are invovled - namely it calles getattr before checking for a slot of the same name, in an alternating fashion
|
|
method now yields good results on all tested platforms
|
|
using a / in a hardcoded fashion, leading to absolute paths where the caller expected relative ones
|
|
it. Its can be assumed though that there are more bugs related to unicode hanging around in the system
|
|
|
|
a few more resT syntax errors on the way
|
|
The first one is faster, although I would have expected the latter one to be faster
|
|
|
|
memory map. Its totally ridiculous, but fixed
|
|
kind of issue doesn't popup for anyone
|
