| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
A non-zero exit status from 'git push' doesn't necessarily mean that
nothing was pushed - we should parse the output just as for a zero
exit status. This requires adding a new option _ignore_exit_status
to git_run().
For --dry, we can actually pay attention to the return code, but need
to fix up previous logic where we returned None instead of a list of
commits, and then tried to iterate over None. For clarity, let
CalledProcessError throw through rather than special-casing a return
of None.
Inspired by a patch from Philip Withnall.
https://bugzilla.gnome.org/show_bug.cgi?id=620668
|
|
|
|
|
|
|
|
| |
If an URL ends up with a unicode type, then that will force the
body of the HTTP request to unicode; adding the raw bytes of the
commit can then cause an encoding error.
https://bugzilla.gnome.org/show_bug.cgi?id=745186
|
|
|
|
|
|
|
| |
Add documentaton of recognized forms of bug references in commit messages,
and explain the way that "See" suppresses bug reference identification.
https://bugzilla.gnome.org/show_bug.cgi?id=708158
|
|
|
|
|
| |
Though it didn't matter here, s.find(x) >= 0 is is better to write
than s.find(x) > 0.
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=673828
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Specify that bugzilla.redhat.com is https - filing still won't work
without specifying a default-version but we'll get a clean error
message since it will now go through the XML-RPC interface
and not be broken by the redirect (see bug 745100)
Since the version on bugzilla.redhat.com is product specific,
and we can't guess what products people will file bugs against,
specifying a default version doesn't make sense.
https://bugzilla.gnome.org/show_bug.cgi?id=673828
|
|
|
|
|
| |
Existing docs made it a bit hard to figure out how --fix works - list
the usage explicitly.
|
|
|
|
|
|
|
| |
bugzilla.gnome.org now supports attachment status on
attachment.cgi?action=insert.
https://bugzilla.gnome.org/show_bug.cgi?id=745062
|
|
|
|
|
|
| |
Handle the changed name of the attachment status field; status on creation
isn't supported with the current b.g.o code, so leave that as a FIXME
for now.
|
|
|
|
|
|
|
| |
create_patch() now gets attachment.cgi to get a token - with the
switch urllib2 this requires slightly different code.
https://bugzilla.gnome.org/show_bug.cgi?id=725730
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
At times, we can make many separate requests. If we start off with
a URL that bugzilla always redirects (say a HTTP URL), it's better to
"get the message" and start using the target of the redirects.
The code that used to do this was removed with the switch to urllib2.
Add back equivalent code - it can be simplified at this point because
bugzilla was changed to use 301 permanent redirects for HTTP => HTTPS
redirects, and left using 302 requests for attachment
redirects.
https://bugzilla.gnome.org/show_bug.cgi?id=725730
|
|
|
|
|
|
|
| |
This simplifies the code and will auto-detect the users proxy
settings and use those.
https://bugzilla.gnome.org/show_bug.cgi?id=725730
|
|
|
|
|
|
| |
So the cache file will now typically be ~/.cache/git-bz-cache.
https://bugzilla.gnome.org/show_bug.cgi?id=674020
|
|
|
|
|
|
|
|
|
|
| |
By treating all strings internally as Unicode (decoding them on input
and encoding them on output), git commit messages which contain Unicode
can be handled without getting UnicodeDecodeErrors.
This works on Python 2 and 3.
https://bugzilla.gnome.org/show_bug.cgi?id=684578
|
|
|
|
|
|
|
| |
Although none of the code or strings need it at the moment, it’s best
to be explicit.
https://bugzilla.gnome.org/show_bug.cgi?id=684578
|
|
|
|
|
|
|
| |
As a follow up to commit 68256107, change the default platform for GNOME
Bugzilla to the empty string, and ignore empty platform fields when
uploading. This makes Bugzilla use the default value, which would be
‘Other’ or ‘All’ or whatever the product has set up.
|
|
|
|
|
|
|
|
| |
It was previously ‘All’, but this seems to have been removed for all
products with the upgrade to Bugzilla 4.4. Change the default to
‘Other’, which is still present.
https://bugzilla.gnome.org/show_bug.cgi?id=744319
|
|
|
|
| |
https://bugzilla.gnome.org/show_bug.cgi?id=721962
|
|
|
|
|
|
| |
At least bugzilla.redhat.com ones are stored like that.
https://bugzilla.gnome.org/show_bug.cgi?id=673826
|
|
|
|
|
|
| |
So that people know where to report git-bz bugs.
https://bugzilla.gnome.org/show_bug.cgi?id=742370
|
|
|
|
|
|
|
| |
Title it ‘AUTHORS’ rather than ‘Author’ to make it more in line with
other man pages.
https://bugzilla.gnome.org/show_bug.cgi?id=742370
|
|
|
|
|
|
|
|
|
|
| |
attachment.cgi requires a token field for Bugzilla 4.2 and later.
(See https://bugzilla.mozilla.org/show_bug.cgi?id=703983)
This token is not the same as the overall bug's token. So from
we need to load attachment.cgi as a GET before every POST, which
is a bit of a pain.
https://bugzilla.gnome.org/show_bug.cgi?id=700502
|
|
|
|
| |
Add $(VPATH), correct paths
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
See http://people.gnome.org/~walters/docs/build-api.txt
This way I can easily add it to the gnome-ostree manifest and have it
install into /usr, package people can do their thing, etc.
This dead simple non-autotools configure/Makefile system is adapted
from one I wrote for gtk-doc-stub.
Note I also added a --disable-documentation flag since gnome-ostree
doesn't have asciidoc.
|
|
|
|
|
|
|
| |
The code that applied plain diff tried to add a Unicode header
to uninterpreted patch data.
https://bugzilla.gnome.org/show_bug.cgi?id=687734
|
|
|
|
| |
Pointed out by Dan Winship
|
|
|
|
|
|
| |
Cookie jar got migrated from ~/.gnome2/epiphany to ~/.config/epiphany
https://bugzilla.gnome.org/show_bug.cgi?id=676241
|
|
|
|
|
|
|
|
|
| |
For plain 'git bz attach', always obsolete existing patches if they
have the same subject as the commit being attached.
For 'git bz attach -e', make this the default.
https://bugzilla.gnome.org/show_bug.cgi?id=671941
|
|
|
|
| |
""" not """"
|
|
|
|
|
| |
Fix a double addition of the commit message, and add a comment
pointing out that the exact string is checked by git.gnome.org.
|
|
|
|
|
|
| |
The author and subject might be wrong (especially the subject), when
we make up a commit message for a plain patch, so ask the user to
double check.
|
|
|
|
|
| |
If the editor uses the terminal, the missing _interactive=True will
cause things to hang.
|
|
|
|
|
|
|
|
|
| |
If a bug has a plain diff (as opposed to git-format-patch output),
prepend some minimal headers to it to get it into a format git-am will
accept, and then make the user write a commit message for it before
proceeding.
https://bugzilla.gnome.org/show_bug.cgi?id=657558
|
|
|
|
|
|
|
|
|
|
|
| |
Rather than failing completely when git-am fails to apply a patch,
write out our current state to git-am's temporary directory, and then
tell the user to use "git bz apply --continue/--skip/--abort" after
handling the merge failure. When the user uses one of those options,
read back our state, pass the flag on to git-am, and (assuming git-am
succeeds), continue with the next patch.
https://bugzilla.gnome.org/show_bug.cgi?id=657558
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Rather than asking whether or not to apply each patch individually,
list all of the patches at once and allow the user to either apply all
of them, apply none of them, or interactively choose which ones to
apply (a la interactive rebasing).
In addition to choosing which patches to apply, the interactive option
also allows reordering the patches before applying, and choosing to
apply committed or rejected patches.
https://bugzilla.gnome.org/show_bug.cgi?id=657558
|
|
|
|
|
|
|
|
|
|
| |
Using 3-way merge makes it much more likely that the patch will apply
correctly when the local tree is newer than the tree that the patch
was generated from, and if the patch doesn't apply cleanly, git will
leave a file with conflict markers rather than forcing you to reapply
the patch manually.
https://bugzilla.gnome.org/show_bug.cgi?id=657558
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The two lines for the different versions of 'git bz edit' are treated
by asciidoc as a single paragraph and line-wrapped.
In theory, a + at the end of a line is a forced line-break with
asciidoc, but in the manpage generation route, this doesn't work
properly - the magic in a2x that is supposed to turn the asciidoc-br
processing instruction into a line break doesn't work. (We aren't
using a2x at the moment for the conversion, but no real reason to
switch if it doesn't work better.)
So, use a [verse] paragraph instead. This results in mis-indentation,
but slightly better than the previous breakage.
https://bugzilla.gnome.org/show_bug.cgi?id=670686
|
| |
|
|
|
|
|
|
|
| |
If we strip leading space from the body, we'll unindent the
first line of the body, which could break aligment.
Based on a patch by Stef Walter <stefw@gnome.org>
|
|
|
|
| |
Pass all the necessary fields to merge_default_fields_from_dict().
|
|
|
|
|
|
| |
Allow setting the default-product in the tracker config. This is possibly
useful for something like bugzilla.redhat.com where many repositories share
a product of "Fedora".
|
|
|
|
|
|
|
| |
Ideally I'd like git-bz to know when a Bugzilla instance supports
these, but here at least we'll try parsing them.
https://bugzilla.gnome.org/show_bug.cgi?id=657361
|
|
|
|
|
|
|
|
|
|
|
| |
If you want to override something like the default version on a per-component
basis, having to set that in the per-tracker configuration is unnatural
and confusing. Allow setting bz.default-version instead. This is interpreted
as overriding the existing configuration of default field values.
Based on a patch from Dan Winship <danw@gnome.org>
https://bugzilla.gnome.org/show_bug.cgi?id=651756
|
|
|
|
|
|
|
| |
Instead of making calls to git.config() as needed, read all of our options
ahead of time into a dictionary initialized with defaults.
Loosely based on a patch from Felipe Contreras <felipe.contreras@gmail.com>.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
All the logic to add the user to the cc: list on the bug is done in the
web page for show_bug.cgi and not on the client side, so the option
"Automatically add me to the CC list of bugs I change" in the user's
preferences is not honored. Since that preference defaults to Always
and users will very typically want to see further traffic on their patches,
just always request the user to be added to the Cc.
The main downside to this is that we do another round trip to the server
when creating and attachment, and also potentially create more bug mail.
https://bugzilla.gnome.org/show_bug.cgi?id=663800
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
If the data that we are writing into the template ends includes
non-ASCII data, it will need to be encoded as UTF-8 before writing
the template file. Fix from Luca Bruno.
https://bugzilla.gnome.org/show_bug.cgi?id=657716
|
|
|
|
|
|
|
| |
If 'git bz file' is used with a commit message that contains a
bug reference, prompt before filing again.
https://bugzilla.gnome.org/show_bug.cgi?id=624461
|