Thu, 19 May 2011 14:02:55 +0000 |
hodgestar |
Allow passing the build master's URL without the /builds suffix when calling bitten-slave by appending /builds to URLs that don't have it. This also prevents bitten-slave from contacting the wrong domain if a bad URL is passed.
|
Wed, 30 Mar 2011 13:29:44 +0000 |
osimons |
Oops. Typo in `local` and `dry_run` slave attachment info logging. Fixes #670. Thanks.
|
Wed, 09 Mar 2011 15:00:49 +0000 |
hodgestar |
Remove second import of os module.
|
Wed, 09 Mar 2011 14:48:35 +0000 |
hodgestar |
Use our own HTTPBasicAuthHandler under Python 2.6 to avoid issue http://bugs.python.org/issue8797. Fixes #658.
|
Sat, 05 Mar 2011 02:15:03 +0000 |
osimons |
Set 'Content-Length' header on keep-alive requests. Fixes #667. Thanks eblot.
|
Fri, 21 Jan 2011 08:13:15 +0000 |
osimons |
Make sure slave reads attachments as binary files. Fixes #657.
|
Mon, 18 Oct 2010 14:03:20 +0000 |
hodgestar |
Fix docstring complaints from epydoc.
|
Sun, 10 Oct 2010 20:18:53 +0000 |
osimons |
Slave attachment support via `<attach />` is totally redone to use multi-part form post instead of inlining it in the XML (ie. like a web file upload form). For larger binaries the previous inlining would effectively be an internal denial-of-service attack...
|
Wed, 06 Oct 2010 09:49:05 +0000 |
osimons |
Updated copyright to 2010.
|
Fri, 14 May 2010 10:48:04 +0000 |
osimons |
Don't initialize the keep-alive thread when running a local build (it will fail trying to get the opener).
|
Thu, 29 Apr 2010 18:17:05 +0000 |
wbell |
Adding keepalives to the bitten client/server protocol.
|
Sat, 24 Apr 2010 14:28:41 +0000 |
wbell |
Make all times be generated by the server. This eliminates a number of inconsistencies you'll see when the clocks on the slaves are skewed from the master.
|
Sat, 24 Apr 2010 13:37:26 +0000 |
wbell |
Adding the ability to modify the default ''onerror'' property in the ''<build>'' element. If not specified, the behavior is unchanged; by default any step failure will result in the build failing and stopping.
|
Wed, 09 Dec 2009 11:37:58 +0000 |
osimons |
Updating protocol version counter following change in [793] / #459, and also improved errors so that the reason is output on console without verbose logging enabled. Fixes #511.
|
Wed, 25 Nov 2009 08:12:04 +0000 |
osimons |
Add a `referer=` argument when logging in, and redirect to a Bitten URL so that slave doesn't need non-build permissions. Closes #459.
|
Thu, 10 Sep 2009 06:54:20 +0000 |
osimons |
0.6dev: Adding some error-handling to bitten-slave for config files. It now reports errors and exists gracefully if a) config file isn't found, or b) problem parsing content (wrong format).
|
Tue, 25 Aug 2009 09:09:26 +0000 |
osimons |
0.6dev: Moved a constant declaration so that master should still work with Python 2.4. Updated docs with regards to Python requirements + added a changelog message from cmlenz.
|
Tue, 25 Aug 2009 00:16:58 +0000 |
osimons |
0.6dev: Removing code and updated docs related to Trac < 0.11 and Python < 2.4 (base requirements for Bitten 0.6).
|
Mon, 24 Aug 2009 22:41:08 +0000 |
osimons |
0.6dev: Switching to use the new cookie-support, and using trac auth/session ID as unique identification for linking builds with slaves. This overcomes problems with IP address not being unique behind NAT, and also where IP address may change during a build. Closes #421.
|
Mon, 24 Aug 2009 22:33:26 +0000 |
osimons |
0.6dev: Adding real authentication by making a request to `/login` and storing and reusing cookies provided by Trac (#208). Additionally, support for account-manager form authentication is added to the slave (#219).
|
Mon, 24 Aug 2009 13:19:27 +0000 |
osimons |
0.6dev: Follow-up to [719]. Seems some http errors don't populate `headers` (at least apache digest auth challenge didn't).
|
Mon, 24 Aug 2009 12:00:43 +0000 |
osimons |
0.6dev: Switching `master.py` to use new `self._send_response()` and `self._send_error() methods. Simplifies code, but most importantly for errors it allows a consistent method for transmitting plain-text error messages to the slave (that the slave will now output as part of debug logging).
|
Tue, 11 Aug 2009 22:05:53 +0000 |
osimons |
0.6dev: Implement a `subprocess` implementation of `execute()` that is now the default for all slaves where this is available - essentially all slaves with Python 2.4 and higher, or where module is installed separately. This fixes:
|
Sat, 01 Aug 2009 01:43:56 +0000 |
osimons |
0.6dev: Logging start + stop time for the slave. Closes #268.
|
Sat, 25 Jul 2009 00:37:23 +0000 |
osimons |
0.6dev: Adding `-P` option to bitten-slave to prompt for password as alternative to command-line option or configuration file. Closes #402.
|
Sun, 12 Jul 2009 21:51:23 +0000 |
osimons |
0.6dev: In the event that loading Recipe fails, `basedir` would not yet have been defined, and removing it would fail.
|
Tue, 07 Jul 2009 15:10:29 +0000 |
osimons |
0.6dev: Fixing #229 again - keeping `slave.work_dir` if either `--keep_files` or `--work_dir=` are part of options.
|
Sun, 05 Jul 2009 11:41:52 +0000 |
osimons |
0.6dev: Fixing keep_files (`-k`) option, closing #229.
|
Sun, 05 Jul 2009 11:33:09 +0000 |
osimons |
0.6dev: Fixing digest authentication in #330. Instead of having a single `urllib2` opener for the life of the slave, a new opener object is created for each request.
|
Mon, 11 May 2009 11:16:15 +0000 |
dfraser |
Fixed up misapplied patch from r633 to have real base class for error processor (see #163)
|
Mon, 11 May 2009 11:01:34 +0000 |
dfraser |
Apply patch to allow Python 2.3 compatibility (remy blank) - fixes #163
|
Fri, 03 Apr 2009 07:41:28 +0000 |
dfraser |
Only actually run `rmtree` if the root directory exists (fixes #381; patch is from there)
|
Tue, 31 Mar 2009 17:25:40 +0000 |
dfraser |
Allow authentication credentials to be specified in the configuration file `[Remy Blank]`.
|
Fri, 27 Mar 2009 10:43:12 +0000 |
dfraser |
Pass `--no-loop` option to the `BuildSlave` (fixes #375, from the patch there - thanks!)
|
Mon, 23 Mar 2009 19:19:17 +0000 |
wbell |
Small documentation fixes.
|
Tue, 17 Mar 2009 06:53:58 +0000 |
dfraser |
Added option for not looping after check. Fixes #371
|
Wed, 11 Mar 2009 12:52:44 +0000 |
dfraser |
Added missing `sys` import - fixes #345
|
Mon, 14 Jul 2008 14:49:19 +0000 |
dfraser |
`os.EX_OK` etc are only defined on UNIX and Mac. Replace with fallback constants. Fixes #298 so Windows slaves now run successfully
|
Tue, 08 Jul 2008 15:17:16 +0000 |
cmlenz |
Apply patch by Emmanuel Blot to allow a single slave to connect to multiple masters. Closes #271.
|
Wed, 18 Jun 2008 18:19:35 +0000 |
mgood |
make `bitten-slave` return a meaningful exit code
|
Fri, 14 Mar 2008 20:57:08 +0000 |
wbell |
Slaves should continue to run builds without sleeping as long as new builds are available. Applied patch from Thomas Mueller. Closes #181
|
Wed, 05 Mar 2008 20:43:55 +0000 |
wbell |
Committing rmtree patch for windows submitted by Doug Patterson. Thanks! Closes #211, #183, #236.
|
Tue, 23 Oct 2007 15:36:06 +0000 |
cmlenz |
Applied patch to #188 for stable/configurable names of build directories. Thanks to Allen Bierbaum for the patch.
|
Wed, 17 Oct 2007 23:10:10 +0000 |
mgood |
handle HTTPErrors separately to work around them not having a "reason" attribute like URLError
|
Mon, 24 Sep 2007 12:56:41 +0000 |
cmlenz |
When the slave is run in `--dry-run` mode, it now also sends a build cancellation message to the master in case of failed builds. Closes #187.
|
Tue, 18 Sep 2007 12:18:34 +0000 |
jonas |
Make the slave differentiate between benign and fatal network errors and
|
Wed, 12 Sep 2007 19:42:11 +0000 |
cmlenz |
Add debug log message for enabling authentication.
|
Thu, 30 Aug 2007 08:07:22 +0000 |
cmlenz |
Fix default for poll interval of slave.
|
Thu, 23 Aug 2007 19:41:22 +0000 |
cmlenz |
Make the interval the slave sleeps between requesting builds configurable from the commandline.
|
Tue, 21 Aug 2007 16:42:53 +0000 |
cmlenz |
Improve the still experimental support for using `coverage.py`.
|
Fri, 17 Aug 2007 10:43:09 +0000 |
cmlenz |
Finish the move of build configuration management into the admin interface.
|
Sat, 11 Aug 2007 19:20:27 +0000 |
cmlenz |
Rudimentary implementation of ``dry-run`` option in build slave. Builds are still put in in progress mode, but no results are transmitted back, and the build is automatically invalidated after the slave is done.
|
Wed, 08 Aug 2007 17:53:18 +0000 |
cmlenz |
Slaves now attempt to explicitly cancel builds when they are interrupted.
|
Wed, 08 Aug 2007 13:10:58 +0000 |
cmlenz |
Fix build slave error introduced in [461].
|
Wed, 08 Aug 2007 13:09:32 +0000 |
cmlenz |
Change the default log verbosity of the build slave, and remove the `--debug` option.
|
Wed, 08 Aug 2007 12:10:46 +0000 |
cmlenz |
The build slave can now run locally against a recipe file, which is useful for testing recipes. Simply pass the path to the recipe instead of the URL of the build master to the script.
|
Tue, 07 Aug 2007 11:29:11 +0000 |
cmlenz |
Use reStructuredText as the API documentation syntax.
|
Tue, 07 Aug 2007 08:58:38 +0000 |
cmlenz |
Update file headers and other stuff pointing to the old home.
|
Fri, 03 Aug 2007 16:37:51 +0000 |
cmlenz |
The slave logs each request at DEBUG level.
|
Fri, 03 Aug 2007 14:44:29 +0000 |
cmlenz |
Switch to urllib2 in slave so that we can support basic and digest authentication.
|