| Commit message (Collapse) | Author | Age | Files | Lines |
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This functionality is now better fulfilled by using an off-the-shelf network
file system technology to share the BE object directory. This has the advantage
of a more comprehensive and integrated security model as well.
|
| |
|
|\
| |
| |
| |
| | |
Fix up some missing changes from the removal of --auth
See merge request bugseverywhere/bugseverywhere!9
|
| |
| |
| |
| |
| | |
The --auth flag was removed in 5fb6a912cd7cb0bcfa4512da5248baad1175faf5 that
should have removed this text as well.
|
| |
| |
| |
| |
| | |
This should have been removed in 5fb6a912cd7cb0bcfa4512da5248baad1175faf5 as it
was related to the preceding sentence that was removed.
|
|/ |
|
|
|
|
|
|
| |
AuthenticationApp (just removed) was the only code that had any interaction with
this functionality. That is, check_login looked for an environment variable
"be-auth.user" that was only ever set by AuthenticationApp.
|
| |
|
| |
|
|
|
|
|
| |
The implementation of this option contained syntax errors and did not work. For
more information, see https://gitlab.com/bugseverywhere/bugseverywhere/issues/7.
|
| |
|
| |
|
|
|
|
| |
Arch was the only VCS that needed this peculiarity.
|
|
|
|
|
|
| |
The last release of GNU Arch was in 2006, over ten years ago at time of writing.
GNU suggests users should migrate repositories to Bazaar. This commit removes
all support for Arch to reduce ongoing maintenance overheads.
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
The calling below seems like a typo to me. How can we index with
a variable which has not been initialized yet?
|
|
|
|
| |
Signed-off-by: Matěj Cepl <mcepl@redhat.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The old implementation just printed the bug UUID (without the bugdir/
prefix). This lead to the command we suggest in `be target --help`:
$ be depend --status -closed,fixed,wontfix --severity -target \
$(be target --resolve)
failing with an invalid ID.
Reported-by: Michael Sperber <sperber@deinprogramm.de>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Comment nesting wasn't working properly before, where you could get
things like:
<div class="comment root" id="a">
</div> <!-- close a -->
<div class="comment" id="a/b">
<div class="comment" id="a/b/c">
<div class="comment root" id="d">
</div> <!-- close d -->
</div> <!-- close a/b/c -->
</div> <!-- close a/b -->
from a comment tree (using fake ids) of:
.
|-- a
| `-- a/b
| `-- a/b/c
`-- d
The new handling pushes the `div_close(depth)` call to the front of
the comment block, because a comment's depth tells us how many of the
already-rendered comments we need to close. Closing comments at the
top of the block means that we'll always have at least one unclosed
comment to close after the comment loop completes. With the new
handling, we'll get a more appropriate:
<div class="comment root" id="a">
<div class="comment" id="a/b">
<div class="comment" id="a/b/c">
</div> <!-- close a/b/c -->
</div> <!-- close a/b -->
</div> <!-- close a -->
<div class="comment root" id="C29a03522-ed6e-4a9a-8823-23a1c513865f">
</div> <!-- close d -->
Reported-by: Owen Jacobson <owen.jacobson@grimoire.ca>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This bug was introduced in:
commit 400c9251eedbd1caa3b28135b4bc5d2bf2124ac5
Author: W. Trevor King <wking@tremily.us>
Date: Thu Jan 24 02:43:53 2013 -0500
util:wsgi: If we're logging to a file, shift libbe.LOG too
where the removed handlers clobbered our initial
TimedRotatingFileHandler.
|
|
|
|
| |
This avoids attempts to log to stderr if we're daemonizing the server.
|
|
|
|
|
|
|
| |
This makes it easier to tweak log verbosity and redirect logs to other
handlers. For example, the WSGI servers are unstable with stderr
closed, and crash with an IOError if they try to print a warning to
stderr.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Not just '\n'. Mark Mikofski reported an error on the upgrade from
1.4 to 1.5 on MS Windows:
> upgrading bugdir from "Bugs Everywhere Directory v1.4" to "Bugs Everywhere Directory v1.5"
> Traceback (most recent call last):
> ...
> File "c:\...\libbe\storage\util\upgrade.py", line 141, in check_initial_version
> assert version == self.initial_version, '%s: %s' % (path, version)
> AssertionError: c:\...\.be\version: Bugs Everywhere Directory v1.4
>
> **notes:** I set a breakpoint and it does seem that they are not the same
> (Pdb) self.initial_version
> 'Bugs Everywhere Directory v1.4'
> (Pdb) version
> u'Bugs Everywhere Directory v1.4\r'
We don't need to convert to Unicode, because on Python 2.7:
$ python2.7 -c "print('a' == u'a')"
True
Strange, but true ;). One day we'll migrate BE to Python 3...
|
|
|
|
|
|
|
|
|
|
|
|
| |
Fix a bug introduced by:
commit 4db1a045a0606bead191a563abc54dfa8352efe0
Author: W. Trevor King <wking@tremily.us>
Date: Wed Aug 29 23:26:17 2012 -0400
Rewrite commands to use bugdirs instead of a single bugdir.
Reported-by: James Spencer <james.s.spencer@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Changes:
* Import libraries with an underscore prefix to avoid polluting the
module's public namespace.
* Use a copy `env` to avoid messing with the child process'
environment.
* Pass the tweaked environment on to the PAGER (e.g. for PATH).
* Handle PAGER='' (by not paging).
* Use shlex.split to handle cases like PAGER='less -FRSX' (thanks to
Da_Blitz from pocketnix.org for the bug report and shlex idea).
|
|
|
|
|
|
|
|
|
|
| |
Nathan's post is released under CC BY-SA 3.0 Unported. According the
the FSF, the CC BY-SA 2.0 Generic was not compatible with BE's GPLv2+
[1]. I expect my Python implementation is sufficiently different from
Nathan's original notes to be classified as "fair use" and not an
"adaptation".
[1]: http://www.gnu.org/licenses/license-list.html#ccbysa
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
This brings BEExceptionApp into closer agreement with
libbe.ui.command_line.dispatch(), and avoids having the server go down
when a user submits a silly command.
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This allows you to manage BE servers from inetd scripts, etc.
Shortcomings of the current implementation:
* ServerCommand._daemonize() currently only sets a SIGTERM handler and
double forks. If you want to do this right, see PEP 3143.
Unfortunately, the PEP seems to have stalled, python-daemon appears
unmaintained, and I don't care enough at the moment to do this
right.
* ServerCommand._get_pidfile() races between checking for an existing
PID file and claiming the file itself. It is possible that two
processes would check around the same time, and both see no existing
file. Then they would both open the PID file and write their pid,
without noticing that the other process was contending for the file.
Solving this requires file locking, which is difficult to do
portably. This shouldn't be an issue in normal operation, where
each server will be using its own PID file path.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
This was broken in the creation of libbe.util.wsgi from
libbe.command.serve:
commit 0fa17f6bf6a809df14ae1930542059a9e62066b0
Author: W. Trevor King <wking@tremily.us>
Date: Mon Aug 27 13:37:58 2012 -0400
libbe:util:wsgi: extract WSGI utilities into a separate module.
|
|
|
|
|
|
|
|
|
|
|
| |
This was lost in the creation of libbe.util.wsgi from
libbe.command.serve:
commit 0fa17f6bf6a809df14ae1930542059a9e62066b0
Author: W. Trevor King <wking@tremily.us>
Date: Mon Aug 27 13:37:58 2012 -0400
libbe:util:wsgi: extract WSGI utilities into a separate module.
|
| |
|
| |
|
| |
|