|
From: | Patrick Alken |
Subject: | Re: [Bug-gsl] [Help-gsl] test release 1.99.91 |
Date: | Thu, 29 Oct 2015 09:47:32 -0600 |
User-agent: | Mozilla/5.0 (X11; Linux x86_64; rv:38.0) Gecko/20100101 Thunderbird/38.3.0 |
Patrick On 10/29/2015 05:43 AM, Raymond Rogers wrote:
I suggest a script for validation testing that tries all optimization settings and evaluates them separately; and records the results. The reason is that optimization seems to vary with gcc/compiler version. It's my belief that program maintenance must be scripted in order to record past failings (which unfortunately reoccur) and provide feedback to maintainers. Otherwise there are endless back and forth's with people like myself reporting this and then that; and the other thing. If there is interest I can try to make up such a script/test for GSL in some form for linux. Suggestions for the script basis would be appreciated. I am inclined towards an automatizable interactive language or bash or AWK. Or perhaps there is a "unversal" software test program that allows control, repetition with variation; and most importantly good record keeping on both the input and output. Then the maintainers can just say: run this script and send the results back.Ray On 10/29/2015 06:31 AM, Rhys Ulerich wrote:I have "icc version 14.0.3 (gcc version 4.8.1 compatibility)" here andintel compiles the gsl package ok without any compiler warnings. Thank you. Does make check pass at -O2? 3? If not 2, 1? I'm not concerned about compilation as much as the unit tests passing. - Rhys
[Prev in Thread] | Current Thread | [Next in Thread] |