This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [tree-ssa, LNO] Analysis of scalar evolutions and data dependences
- From: law at redhat dot com
- To: Pop Sébastian <pop at gauvain dot u-strasbg dot fr>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Wed, 17 Dec 2003 09:36:03 -0700
- Subject: Re: [tree-ssa, LNO] Analysis of scalar evolutions and data dependences
- Reply-to: law at redhat dot com
In message <20031217095357.GB24070@gauvain.u-strasbg.fr>, =?iso-8859-1?Q?Pop_S=
E9bastian?= writes:
>On Tue, Dec 16, 2003 at 09:36:06PM -0700, law@redhat.com wrote:
>> I really don't think this is the right way to go because the dumps
>> change if we suddenly use an extra temporary during gimplification,
>> or delete a PHI and recycle an SSA_NAME that wasn't recycled before,
>> or due to addresses of temporary variables changing, etc etc.
>>
>
>If an analyzer does not dump data that is susceptible to change
>independently of its behavior, the diffing of the dump could be a good
>strategy for testing.
Sorry, I should have been more specific. If you're looking at detail
data emitted by the pass rather than the C-like representation of the
function, then this may be a reasonable scheme. If you go that
direction your biggest problem is likely the fact that the C-like dump
files and the dump details for a pass are in the same file.
>> Instead I would ask that you look into how to expand the testing
>> infrastructure to allow you do write tests more reasonably without
>> just diffing the dump files.
>>
>
>Okay, I will try to find a better strategy for testing the analyzers
>outputs.
As you get ideas, do feel free to bounce them off me -- I don't need it
immediately, but I'm on the lookout for a good way to do testing of
code motions.
jeff