Hi Martin,
>>> PS I'm not happy about duplicating the same test across all those
>>> targets. It would be much nicer to have a single test somewhere
>>> in dg.exp #include a target-specific header with macros describing
>>> the target-specific parameters.
>>
>> why so complicated? Just have a single attr-aligned.c test, restricted
>> to the target cpus it supports via dg directives and with
>> MINALIGN/MAXALIGN definitions controlled by appropriate target macros?
>
> I have done that in the past(*) but hardcoding target-specific
> assumptions into a general test didn't feel right either given
> the design of the test suite (separate target tests).
>
> [*] see the tangle of #ifdefs in gcc/testsuite/gcc.dg/attr-aligned.c
however, gcc.target, g++.target are only meant for tests that only make
sense on a particular target and cannot work elsewhere. The current
amount of duplication between the various gcc.target/*/attr-aligned.c
tests (where the only variation I could see are the MINALIGN/MAXALIGN
definitions) seems way worse. For someone adapting the test to his
target, it's way easier to skim through the variations (hopefully with
appropriate comments where non-obvious) in a single file rather than
looking at many almost identical files in gcc.target.
Rainer
--
-----------------------------------------------------------------------------
Rainer Orth, Center for Biotechnology, Bielefeld University
