perf(output): avoid rezipping output for managed studies - #3369
Conversation
MartinBelthle
changed the title
| # - be more explicit | ||
| # - not affect internal studies | ||
| if job_launch_params.archive_output: | ||
| if not is_study_managed: |
There was a problem hiding this comment.
Why this isn't not is_study_managed and job_launch_params.archive_output ? I get that archive_output is depreceated but should still be supported untill it's removed no ?
There was a problem hiding this comment.
Also if we do this if not is_study_managed or not job_launch_params.auto_unzip then we won't need to run an archive on it later, I mean we could get rid of :
if output.is_dir() and not auto_unzip:
# This only happens for managed studies (not obvious due to the leaky workflow)
self.archive_output(uuid, output_id)There was a problem hiding this comment.
The archive_output did not make any sense (always True, non-editable in the front). The only parameter that matters is auto_unzip where if it is False the output is left archived and otherwise it is automatically unarchived. So I deprecated it.
There was a problem hiding this comment.
And for the 2nd comment, the if actually matters as now the output always arrives as a folder inside the study for managed studies (except for the -z option but we do not use it) so if the user did not ask for automatic unzip we have to re-zip it
3 participants
Solves the performance issue inside [ANT-5556] but not the design flaws