[Debichem-devel] Bug#475064: please consider making testsuite run shorter

[Riku Voipio]

Package: apbs
Version: 0.5.1-2
Severity: wishlist

Looking at build-logs of a my pet meager arch (armel):

grep -E '(Testing|elapsed)' apbs_0.5.1-2_20080408-0023
...
Testing ion at -3.000
Total elapsed time: 3040 seconds
Testing ion at -2.750
Total elapsed time: 3028 seconds
Testing ion at -2.500
Total elapsed time: 3029 seconds
Testing ion at -2.250
Total elapsed time: 3026 seconds
Testing ion at -2.000
Total elapsed time: 3045 seconds
Testing ion at -1.750
Total elapsed time: 3027 seconds
Testing ion at -1.50
Total elapsed time: 3024 seconds
Testing ion at -1.250
Total elapsed time: 3026 seconds
Testing ion at -1.00
Total elapsed time: 3056 seconds
Testing ion at -0.750
Total elapsed time: 3025 seconds
Testing ion at -0.500
Total elapsed time: 3022 seconds
Testing ion at -0.250
Total elapsed time: 3027 seconds
Testing ion at 0.000
Total elapsed time: 3035 seconds
Testing ion at 0.250
Total elapsed time: 3024 seconds
Testing ion at 0.500
...

Looking at the actual testfile, it indeed seems
apbs is running the same 1h long test with the
only difference being the initial coordinates.

Unless all the seed coordinates are known
corner cases (which seems unlikely considering
the 0.250 stepping..), it would seem this repeated
testing gives little extra value to testing.. Running
test with three sets (negative, zero, positive) would
likely give same code coverage.


-- 
"rm -rf" only sounds scary if you don't have backups
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
Url : http://lists.alioth.debian.org/pipermail/debichem-devel/attachments/20080408/dbf0f741/attachment.pgp