Page MenuHomePhabricator

[test-suite] Change printf specifiers to capital
ClosedPublic

Authored by amyk on May 24 2021, 9:21 PM.

Details

Summary

This patch changes some of the specifiers in the printf calls within:

MultiSource/Applications/sgefa/sgefa.test
MultiSource/Benchmarks/TSVC/CrossingThresholds-dbl/CrossingThresholds-dbl.test
MultiSource/Benchmarks/TSVC/CrossingThresholds-flt/CrossingThresholds-flt.test

to use the capital E/G specifiers. It also edits the reference files to account for the capital letter output.
This change was made because on AIX, inf is printed as INF, so the reference output and printf calls
are changed to print in capitals to avoid introducing a new reference output for AIX.

Diff Detail

Repository
rT test-suite

Event Timeline

amyk requested review of this revision.May 24 2021, 9:21 PM
amyk created this revision.

LGTM (it has been acknowledged that the AIX output is not as specified by C; see ISO/IEC JTC 1/SC 22/WG 14 document N2290); thanks!

This revision is now accepted and ready to land.May 27 2021, 4:40 PM
amyk updated this revision to Diff 389892.Thu, Nov 25, 6:37 PM
amyk edited the summary of this revision. (Show Details)

Updated the revision as upon testing prior to committing, I realized that this patch originally meant to use the capital
E/G printf specifiers, rather than E/F as I originally had.

The reason being is that the pertinent parts of the CrossingThresholds reference output that needs to be capitalized is
the INF of Checksum because printing infinites on Linux is lowercase by default, whereas it is always uppercase on AIX.
The updates to the capital G printf specifier in MultiSource/Benchmarks/TSVC/tsc.inc is the actual the printf that
controls the Checksum column in the reference output (and not the F I updated in the original patch).

I must have forgotten to test Linux when I posted the patch, as I remember testing this patch and having it pass
(so it must have been on AIX). This current update passes both Linux and AIX as far as my testing goes.

@hubert.reinterpretcast If you could take another quick look at MultiSource/Benchmarks/TSVC/tsc.inc, that would be greatly appreciated.

daltenty accepted this revision as: daltenty.Fri, Nov 26, 7:00 AM

LGTM, thanks

This revision was automatically updated to reflect the committed changes.