| Age | Commit message (Collapse) | Author |
|---|
|
|
|
It shows that the previous implementation was never really working on
linux, and thus failed on travis as well for good reason.
Closes #303
|
|
When expanding directories, check if it is a symlink and don't expand
them at all.
Previously, we followed symlinks and expanded their contents, which
could lead to weird index files.
Fixes #302
|
|
This will make usage of the `NullHandler` possible in python 2.6
and below.
|
|
When the code is run without setting up loggers, the loggers have no
handlers for the emitted messages. The logging module displays:
`No handlers could be found for logger "git.cmd"` on the
console. By adding a NullHandler (a no-op) the message disappears,
and doesn't affect logging when other handlers are configured.
|
|
support passing odbt for using with Repo
|
|
It turns out we can't deal do fetches if no refspec is set as git
will change the format of the fetch return values, providing less
information than usual.
A test was added to show that such a case will fail, and an assertion
will assure we don't attempt to fetch/pull if there is no refspec
for 'fetch'.
Closes #296
|
|
In that case, the handler for processing stdout and stderr of the git
process is offloaded to threads. These currently don't return
any exception they raise.
We could easily fix this using an approach as shown
[here](http://goo.gl/hnVax6).
|
|
Reverted changes of `fe2fbc5~2`.
This caused `git-pull` to error, which now actually results in a fatal
error while fetching or pulling. Previously we simply didn't check
for this issue.
Now we are back to a `poll` based or threaded concurrent reading from
stdout and stderr to prevent a git process deadlock, and the
aforementioned error.
Related to #297
|
|
victorgp-master
|
|
* renamed `consider_submodules` to `submodules` to be in line with the
existing parameters. Nowadays I would prefer the `consider_` prefix,
but can't change the existing API and thus stick to the current
naming scheme.
* reduced amount of code in one portion to make it more maintainable.
Related to #294
|
|
Javex-submodule_fix
|
|
|
|
There are cases when might not consider a directory dirty when a submodule has
changes. Particular case was git-up stashing changes where submodules are
irrelevant.
|
|
|
|
|
|
This is a pre-emptive measure based on http://goo.gl/l74GC8 .
Related to #290
|
|
Previously, only program messages where forced to the C-locale, now
we force the entire program. That way, we should assure a remote
will not provide us with branch information in any other language but
english.
Related to #290
|
|
|
|
* config parser now handles quoted values correctly. This doesn't hamper
multi-line support.
* added regression test to travis to assure we will be warned if we
rewrite and break the user's .gitconfig file
* only rewrite configuration files if we actually called a mutating
method on the writer. Previously it would always rewrite it.
Fixes #285
|
|
add a GIT_PYTHON_TEST_GIT_DAEMON_PORT to set a port other than 9418,
for example for when you already have a daemon running on that port.
|
|
No reason to expose a daemon to all interfaces when it is only used for
tests, which connect to localhost anyway.
I'd love to use localhost here instead, but the git-daemon man page points out:
If IPv6 is not supported, then --listen=hostname is also not
supported and --listen must be given an IPv4 address.
I don't know of a way to check if git has ipv6 support, but 127.0.0.1
should be around for the foreseeable future
|
|
It expected to see major version 0 though.
|
|
|
|
The previous patch failed to update the expected number of op_codes,
which would result in an exception when creating an instance of
RemoteProgress. This patch corrects the value to the new expected number
of op_codes (8)
|
|
When running a long running operation (such as a clone on a large repo),
Git may return a message indicating that it is 'Finding sources'. Since
there is no bit field value for this message, this causes a large amount
of error messages to be emitted to stderr.
This patch fixes this by adding another bit field value for this
message, FINDING_SOURCES. Derived classes can look for this op_code and
handle it appropriately.
|
|
|
|
Related to #248
|
|
... if it is not found. Previously, especially on windows, this wasn't
explicit.
Fixes #248, affects #126
|
|
Do not swallow non-zero exit status during push and fetch unless
we managed to parse head information.
This behaviour will effetively handle cases were no work was done
due to invalid refspecs or insufficient permissions.
Fixes #271
|
|
When pushing/pulling, we ignore errors unless it's exit code 128.
The reason for this is now made explicit to make clear that issues
are handled by PushInfo flags accordingly.
Related #271
|
|
This should prevent a resource warning given in py3
Fixes #263
|
|
It turned out that the index is not actually corrupted, which is good
news. What happens is that `git` writes `TREE` extension data into the
index, which causes it to write out the given tree *as is* next time
a `git commit` is executed. When using `git add`, this extension data
is maintained automatically. However, GitPython doesn't do that ... .
Usually this is no problem at all, as you are supposed to use
`IndexFile.commit(...)` along with `IndexFile.add(...)`.
Thanks to a shortcoming in the GitPython API, the index was
automatically written out whenever files have been added, without
providing control over whether or not *extension data* will be
written along with it.
My fix consists of an additional flag in `IndexFile.add(...)`, which
causes extension data not to be written by default, so commits can be
safely done via `git commit` or `IndexFile.commit(...)`.
However, this might introduce new subtle bugs in case someone is
relying on extension data to be written. As this can be controlled
through the said flag though, a fix is easily done in that case.
Fixes #265
|
|
Previously it would always adjust your system path, which is bad
behaviour.
|
|
However, it doesn't reproduce on the latest version of GitPython.
Maybe it's on an older one.
|
|
On Windows, `repo.create_submodule(...)` failed because git didn't recognize
the worktree path set in `.git/modules/sub/config` (`fatal: bad config file
line 6 in ./config`). This commit makes `_write_git_file_and_module_config`
convert the worktree path to the linux format (forward slashes) which git
recognizes.
|
|
|
|
If a file in a commit contains no changes (for example, if only the
file mode is changed) there will be no blob attached. This is
usually where the filename is stored, so without it, the calling
context can not tell what file was changed. Instead, always
store a_path and b_path on the Diff object so that information
is available.
|
|
In repositories like
> git branch -a
* test
> ls
test
`repo.iter_commits` failed due to an ambigous argument
(`'git rev-list test`).
Now this cannot happen anymore.
fixes #264
|
|
Related to #224
[ci skip]
|
|
Think about how expensive this single invisible character was, with
all the time and energy spent on it !
|
|
Fixes #260
|
|
While at it, all other invocations of .git in remote.py were reviewed
Fixes #262
|
|
|
|
As I am pretty sure to have tested it with 1.7.0, I assume they
added the git file feature somewhere between .0 .10.
Fixes #252
|
|
Related to #252
|
|
It's new in the latest version of flake - thanks travis for letting me know.
|
|
That, over time, could have caused slow downs due to file-system hassle.
Fixes #258
|
|
Also move untestable documentation out of test.
Related: #234, #242
|
|
Related to #255
|
