Commit graph

14 commits

Author SHA1 Message Date
29edc33337 Coding style. 2016-10-21 21:55:24 +02:00
4fa5b524d4 Coding style.
Using asyncio coroutines for workers instead of threads.
2016-03-09 10:47:09 +00:00
4919a1ed2a Coding style.
More consistent leading underscores.
2016-02-15 14:11:08 +00:00
2373e78cbd Coding style.
Moved the Result class into tools, which simplifies worker.py and makes the
circular import problem go away.
2016-02-13 18:48:53 +00:00
44949c6cf9 Added a command-line option to control the number of workers. 2016-01-31 23:19:28 +00:00
77012c92e4 Make the window in help's golden file bigger, now that theres more help. 2016-01-29 21:30:07 +00:00
678dc89cc9 Changed main integration test to wait for both jobs to be completed before shutting down. 2016-01-28 08:43:44 +00:00
20a2f2d401 Not hiding the special files in the cache.
They are still distinguished from the result files because their names don't
contain a hyphen, which all the result files have.
2016-01-26 15:15:05 +00:00
17516cd831 Fixed bug that caused vigil not to restart if the root directory was renamed. 2016-01-26 14:44:42 +00:00
9b1bfc5dd0 Also send log messages to a real log file, at .vigil/.log. 2016-01-26 10:55:18 +00:00
721cc28d03 Run all tools in a filesystem sandbox 2016-01-21 23:22:42 +00:00
5b894d85b0 Update copyright dates 2016-01-01 17:06:28 +00:00
667dd8ffa5 Run all jobs inside long lived worker processes
- There is less forking, once per worker process, instead of once per job.
- Nice and ionice are applied to the worker processes and hence all jobs run
  by them.
- This allows jobs to benefit from caches filled by previous jobs. For example
  many tools applied to a python file need to calculate if its python2 or python3.
2015-12-31 15:07:30 +00:00
fc4b2ced58 Initial commit 2015-12-14 18:03:11 +00:00