[llvm-commits] llvm-gcc4: nested function support (w/o trampolines)

[Reid Spencer]

Hi Duncan,

On Tue, 2007-02-06 at 10:18 +0100, Duncan Sands wrote:
> A better testcase.  This doesn't mean it's good, just better :)
> The gcc testsuite contains more comprehensive tests.  I thought
> of importing some of them into the LLVM testsuite, but I've decided
> to wait until there is some kind of consensus as to what should be
> done about the gcc testsuite in general - import the relevant bits
> into LLVM?  Turn off the irrelevant bits in gcc?

We discussed this a bit on IRC as Anton reported 10 or so failures in
the gcc.c-torture/compile test suite. llvm-gcc should be able to pass
the test suite, of course. However, we don't want to replicate the
entire gcc test suite in LLVM. What we decided is that if the
failure/assertion/crash occurs in LLVM code while llvm-gcc is running
then we should reduce that test case to the problem and add it in the
appropriate llvm/test sub-directory. In some cases that will be the C
code in CFrontend, in others we can reduce to llvm assembly.  The point
is, we only want to add regressions from regular gcc and only those that
are LLVM's fault. If the failure occurs in llvm-gcc code then we
shouldn't add it to the llvm test suite.

> 
> Ciao,
> 
> Duncan.
> _______________________________________________
> llvm-commits mailing list
> llvm-commits at cs.uiuc.edu
> http://lists.cs.uiuc.edu/mailman/listinfo/llvm-commits