- Add a travis workers with SLURM. By Arnaud Joly, Konstantin Petrov, Olivier Grisel and Loic Esteve
- Add a travis workers with SGE. By Arnaud Joly
- Fix possible deadlock occurence due to the use
subprocess.PIPE
. By Arnaud Joly- Add the possibility to specify a user in
scheduler.queued_or_running_jobs()
. By Arnaud Joly- Automate the detection of the scheduler based on the commands available in the
PATH
withbackend='auto'
(enabled by default). By Olivier Grisel- Make
scheduler.queued_or_running_jobs()
return decoded job names (unicode strings instead of byte strings). By Olivier Grisel- Whenever the
log_directory
argument ofscheduler.submit()
is specified, the job log is now writtend at the location of the log directory under the namejob_name.job_id.txt
wherejob_name
is the name of the jobs andjob_id
is the job id given by the scheduler. The stdout and stderr streams are now joined on SGE (as for the default behavior of SLURM). By Olivier Grisel- Add the possibility to overwrite a value associated to a key already present in the database in
storage.sqlite3_dumps()
instead of raising an IntegrityError. By `Jean Michel Begon`_
- Add support for landscape.io to track code quality. By Arnaud Joly
- Add support for coveralls to track code coverage. By Arnaud Joly
- Add a user guide. By Arnaud Joly
- Add a travis bot for continous integration. By Arnaud Joly
- Add the possibility to get all key-value pairs from the sqlite3 database with
storage.sqlite3_loads()
by settingkey
toNone
. By Arnaud Joly- Add a
storage.sqlite3_loads()
andstorage.sqlite3_dumps()
to cache easily results of functions or scripts in distributed environments. By Arnaud Joly- Add a
scheduler.queued_or_running_jobs()
function to get the names of all running jobs on SGE or SLURM cluster. By Arnaud Joly- Add a
scheduler.submit()
function to write easily job submission query to sun grid engines (SGE) and SLURM. By Arnaud Joly