[MITgcm-devel] testreport - ed can you address this?

[Ed Hill]

On Sun, 14 Jan 2007 14:21:36 -0500 (EST) Andrea Molod
<molod at ocean.mit.edu> wrote:

> hi there,
> 
> i have a comment and a suggestion about testreport.
> at present, when a test (on faulks) fails the monitor comparison
> the output is lost (except for the adjoint tests). this means
> that the test has to be re-run in order to see what went wrong.
> i understand that keeping the output.txt files for all tests
> on the web 'forever' puts a silly strain on disk space. so -
> can we keep the 'last' output.txt from a run that failed the
> monitor comparison (either because it croaked or because there
> is no monitor output) on faulks in /scratch and NOT include
> it in the tar file that gets made to put stuff on the web?
> i say all this for faulks because i am not sure how stuff
> works for remote machines (ie, columbia), and because faulks
> is the machine that is used for the benchmark (so i at least
> am more likely to want to track down why a test there fails....).
> 
> i guess that my main question is whether others would like to
> see this stuff saved, and how much of a pain it would be to
> change testreport and the run scripts that invoke it.....


Hi Andrea,

I checked in a new version that does the above by creating a few
separate sub-directories -- one for each "run".  I've run it and it 
seems to work OK.  If I've inadvertently introduced problems then I 
can sort it out later this week.

Ed

-- 
Edward H. Hill III, PhD  |  ed at eh3.com  |  http://eh3.com/
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://mitgcm.org/pipermail/mitgcm-devel/attachments/20070114/42c04b6c/attachment.sig>