[PATCH] fpcmp fix (wrt sqlite3)

[Renato Golin]

On 5 April 2013 00:47, Michael Gottesman <mgottesman at apple.com> wrote:

> If you really want to test fpcmp, doing all 16 bit float comparisons could
> be accomplished in a reasonable amount of time.
>

I don't, that's the point.

I don't think it's worth doing it now, just before the release, with so
much to do regarding buildbots, tests and benchmarks.

This could be a good project for a summer student, though.

There is a simpler fix, though, to roll back inside BackupNumber as Daniel
suggested, *only* in case the current char is not a number but the previous
is. I'm running some simple cases and the new binary is handling them all
well (while the old one is breaking on every single one of them).

I'll share my results once they're done, and while that, the test-suite is
running with the new fpcmp.

cheers,
--renato
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20130405/aa913e9e/attachment.html>