| Wed, 28 Sep 2005 15:49:27 +0000 |
cmlenz |
* Cleanup and documentation for the `BuildQueue` class added in [236].
|
| Tue, 27 Sep 2005 23:14:48 +0000 |
cmlenz |
* Factor build queue logic into a class separate from the build master.
|
| Tue, 27 Sep 2005 17:05:50 +0000 |
cmlenz |
Really fix BEEP interop, the changes in [233] were broken. There should be no CRLF between payload and trailer.
|
| Tue, 27 Sep 2005 15:40:31 +0000 |
cmlenz |
Remove debugging leftover from [233].
|
| Tue, 27 Sep 2005 15:31:51 +0000 |
cmlenz |
Some BEEP interoperability fixes:
|
| Mon, 26 Sep 2005 22:49:15 +0000 |
cmlenz |
Fix upgrade for installs with no stored reports. Thanks to Matt Good for catching this.
|
| Mon, 26 Sep 2005 17:57:42 +0000 |
cmlenz |
Fix test failure on windows, where a temporary directory could not be deleted due to insufficient permissions.
|
| Sat, 24 Sep 2005 12:21:34 +0000 |
cmlenz |
* Get rid of `xmlio.SubElement`.
|
| Thu, 22 Sep 2005 23:05:00 +0000 |
cmlenz |
Some unit tests for the `bitten.util.archive` module.
|
| Wed, 21 Sep 2005 23:06:12 +0000 |
cmlenz |
Some pylint-inspired cleanup.
|
| Wed, 21 Sep 2005 22:46:15 +0000 |
cmlenz |
Fix regression in line counting in `<python:trace>` command, introduced in [226]. The number of covered lines was mistakenly being used as the total number of lines of code.
|
| Wed, 21 Sep 2005 22:20:11 +0000 |
cmlenz |
Handle relative file paths in `trace.py` output. Closes #51.
|
| Wed, 21 Sep 2005 21:19:42 +0000 |
cmlenz |
Fix error handling when unpacking ZIP archives (see #52).
|
| Tue, 20 Sep 2005 23:10:15 +0000 |
cmlenz |
Fix paging on build config page: most recent changeset was being skipped.
|
| Tue, 20 Sep 2005 23:03:03 +0000 |
cmlenz |
Paging on build configuration page. Closes #24.
|
| Tue, 20 Sep 2005 22:16:41 +0000 |
cmlenz |
* Refactoring of build recipes, the file format has changed slightly:
|
| Tue, 20 Sep 2005 17:19:31 +0000 |
cmlenz |
Removed the use of the `fcntl` module to make the file IO non-blocking in the `CommandLine` class. Instead, use the functions `os.read()` and `os.write()`, which should only block when no data is available -- and that cannot happen because of the preceding `select()`.
|
| Tue, 20 Sep 2005 16:31:03 +0000 |
cmlenz |
Follow-up to [219]: Input to processes now also implemented for Windows.
|
| Tue, 20 Sep 2005 12:02:38 +0000 |
cmlenz |
Implement providing input data to processes executed via the `CommandLine` class. The `<sh:pipe>` recipe command should now be functional. Closes #34.
|
| Mon, 19 Sep 2005 23:41:22 +0000 |
cmlenz |
Hide deactivated build configurations on the build status page by default, and provide an option panel to make them visible.
|
| Mon, 19 Sep 2005 23:28:38 +0000 |
cmlenz |
* Slave now removes build working directories when done.
|
| Mon, 19 Sep 2005 20:30:13 +0000 |
cmlenz |
Remove the wrapper scripts (now taken care of by setuptools/easy_install) and update the docs to reflect the name change.
|
| Mon, 19 Sep 2005 20:13:35 +0000 |
cmlenz |
Ignore the `egg-info` directory created by setuptools.
|
| Mon, 19 Sep 2005 20:09:30 +0000 |
cmlenz |
Make use of the entry point feature introduced in setuptools 0.6a1, and the `console_scripts` entry point introduced in 0.6a2. This also requires Trac >= [http://projects.edgewall.com/trac/changeset/2247 r2247] (or 0.9b2 once that has been released).
|
| Mon, 19 Sep 2005 17:56:29 +0000 |
cmlenz |
* Make deletion of build steps compatible with PySQLite2: records cannot be deleted inside a `select()` loop.
|
| Mon, 19 Sep 2005 15:22:14 +0000 |
cmlenz |
Huge refactoring to remove dependency on BDB XML. Report data is now stored in the Trac database (SQLite/PostgreSQL).
|
| Wed, 14 Sep 2005 19:39:53 +0000 |
cmlenz |
Fix another glitch introduced in [209].
|
| Wed, 14 Sep 2005 19:23:43 +0000 |
cmlenz |
Fix report deletion broken in [209].
|
| Wed, 14 Sep 2005 18:55:51 +0000 |
cmlenz |
Changes to the BDB XML report store to support transactions. Closes #47.
|
| Wed, 14 Sep 2005 18:52:12 +0000 |
cmlenz |
* Improve error handling in the snapshot archive packing/unpacking code. Specificially, when a build slave fails to unpack a snapshot it received from the master, it'll report the error back to the master (and thereby reject the build request), instead of just bailing and leaving the master waiting for a reply.
|
