Bill Deegan
2018-04-05 17:34:26 UTC
Chris,
While it does sound like an interesting application, it may be there
already exist better tools for this.
Are you using Grid Engine or LSF?
Or Celery.
Those would be the first tools I'd reach for if I'm understanding what
you're asking for properly.
In general we discuss use models etc in the users mailing list and shift to
the dev mailing list for details of developing SCons.
Since your email straddles both, perhaps it makes sense to include both
lists, so I'm cc'ing users mailing list.
-Bill
SCons Project Co-Manager
On Thu, Apr 5, 2018 at 9:18 AM, Akins, Christopher (GE Aviation, Non-GE,
While it does sound like an interesting application, it may be there
already exist better tools for this.
Are you using Grid Engine or LSF?
Or Celery.
Those would be the first tools I'd reach for if I'm understanding what
you're asking for properly.
In general we discuss use models etc in the users mailing list and shift to
the dev mailing list for details of developing SCons.
Since your email straddles both, perhaps it makes sense to include both
lists, so I'm cc'ing users mailing list.
-Bill
SCons Project Co-Manager
On Thu, Apr 5, 2018 at 9:18 AM, Akins, Christopher (GE Aviation, Non-GE,
Hey all,
I work for GE Aviation, engine controls group, tools subgroup. Our
application choreographs a lot of steps, from checking out a version of the
controller software, to generating reports on its architecture, to
simulating the controller handling a simulated engine, to churning out
thousands of charts and graphs and binding them all together into immense
PDFs for distribution to the FAA.
These are all done using ad-hoc Perl scripts. Run on one core. Itâs a
travesty.
Iâm trying to modernize the process, and SCons has come favorably to my
attention. Due to the lengthy and memory-intensive simulations, though, it
would be useful if we could use the same build system to assign tasks to
different nodes in a cluster. Iâm willing to sink some time into making the
necessary mods myself â for fault-tolerance, retries, etc â but I thought
Iâd ask if this had come up in the past. If it hasnât, would there be
interest in having a distributed feature, or would it be more of a private
fork?
Thanks,
Chris Akins
_______________________________________________
Scons-dev mailing list
https://pairlist2.pair.net/mailman/listinfo/scons-dev
I work for GE Aviation, engine controls group, tools subgroup. Our
application choreographs a lot of steps, from checking out a version of the
controller software, to generating reports on its architecture, to
simulating the controller handling a simulated engine, to churning out
thousands of charts and graphs and binding them all together into immense
PDFs for distribution to the FAA.
These are all done using ad-hoc Perl scripts. Run on one core. Itâs a
travesty.
Iâm trying to modernize the process, and SCons has come favorably to my
attention. Due to the lengthy and memory-intensive simulations, though, it
would be useful if we could use the same build system to assign tasks to
different nodes in a cluster. Iâm willing to sink some time into making the
necessary mods myself â for fault-tolerance, retries, etc â but I thought
Iâd ask if this had come up in the past. If it hasnât, would there be
interest in having a distributed feature, or would it be more of a private
fork?
Thanks,
Chris Akins
_______________________________________________
Scons-dev mailing list
https://pairlist2.pair.net/mailman/listinfo/scons-dev