Hi,
On Tue, May 28, 2013 at 2:28 PM, <dcerutti.rci.rutgers.edu> wrote:
> Also, I was recently talking with another group that has their own 80-man
> git-based software development project, and their revision control
> includes a lobby for all new code where it is tested for compilation
> compatibility and also runs the test suite before code is then committed
> to the repository. Would it be reasonable to set up something of that
> nature, perhaps with the cheapest GNU compilation as a filter against
> non-portable or broken code, for Amber?
It sounds good in theory, but in practice (at least as far as I
understand what you mean) I think this might be really difficult to
implement. Are you talking about every commit triggering several
compiles (using different compilers), then running a test suite for
each compilation? The current test suite takes a very long time (on
the order of hours) to run, so even if the compiles/tests were run in
parallel, that's still a long time to wait for each commit. What
happens in the meantime if other people make commits?
I like the current system of each user making sure the code compiles
on their machine (serial/parallel), running tests (again
serial/parallel), then making the commit and letting the dashboard
tell you if you broke the compilation using a compiler you may not
have access to.
-Dan
--
-------------------------
Daniel R. Roe, PhD
Department of Medicinal Chemistry
University of Utah
30 South 2000 East, Room 201
Salt Lake City, UT 84112-5820
http://home.chpc.utah.edu/~cheatham/
(801) 587-9652
(801) 585-9119 (Fax)
_______________________________________________
AMBER-Developers mailing list
AMBER-Developers.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber-developers
Received on Tue May 28 2013 - 14:30:03 PDT
