[llvm-commits] [llvm] r80348 - in /llvm/trunk/test/Analysis/Profiling: ./ 2009-08-21-irregular-loop.ll 2009-08-21-only-one-block.ll 2009-08-21-several-blocks.ll dg.exp

[Andreas Neustifter]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi,

Daniel Dunbar wrote:
> Maybe it only makes sense to add the -irregular-loop test; I think its
> very unlikely that the other two will catch errors this one doesn't
> (that couldn't be caught with non-lli testing). 

The 2009-08-21-several-blocks.ll may go, I will look into this.

I would like to keep the 2009-08-21-only-one-block.ll because it tests
the proper implementation of the function entry edge (0,entry) that is
necessary when there is no other edge in the function.

> Also, just in the
> interest of keeping Chris happy, I think really only two calls two lli
> are necessary (to make sure that the code to reload the profile data
> works), any more are superfluous. Does this make sense, or do you
> think these tests will catch other stuff?

I will try to create a subset that has the same test coverage.

> Also, we tend to use the YYYY-MM-DD- format for specific regression
> tests, these can just be named something like
> 'execution-profiling.ll', or whatever best describes the feature they
> are testing.

Okay.

Andi

- --
==========================================================================
This email is signed, for more information see
http://web.student.tuwien.ac.at/~e0325716/gpg.html
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iEYEARECAAYFAkqXh38ACgkQPiYq0rq7s/DU/ACdEC7W7wv4g93ktvR2tFpKalOm
ftsAnRlldyWi+9H9MiTwdnSpuG1RIJ5M
=jZAt
-----END PGP SIGNATURE-----