gfortran test coverage

[Daniel Kraft]

Hi,

I just experimented with running gfortran through gcov (since I can now 
use the compile farm to do experimental builds faster and without 
busying my laptop).  I probably more hacked the build system than did it 
"correctly", but at least it seemed to work basically.  There were some 
test-suite failures probably because the build was not "completely 
sane", but only few.

Nevertheless, I think I managed to get gcov test coverage output for the 
gcc/fortran sources based on the gfortran.dg test-suite.  A quick glance 
over some files shows that there are especially error messages and 
"early return because of special case" code branches that were never 
executed.  Do you think it would be worthwhile to investigate the 
coverage data further and try to extend the test-suite so that coverage 
goes up and/or remove dead code that is really unreachable?  (Or replace 
branches by asserts, for instance.)

I'm somewhat undecided about whether this is something to spend time 
with, but in general I think it would be useful to work on that; this 
hopefully makes gfortran more stable and eases maintainance.  If you 
want, I volunteer to spend some time (I'll probably not have a lot, but 
possibly regularly small chunks) on that during stage 3 (which seems to 
fit for that, also).  And of course I can send the gcov output to anyone 
interested to look at it, too.

Yours,
Daniel

-- 
http://www.pro-vegan.info/
--
Done:  Arc-Bar-Cav-Kni-Ran-Rog-Sam-Tou-Val-Wiz
To go: Hea-Mon-Pri