Re: [Gm2] Debian/Alpha

[Gaius Mulley]

Michael Lambert <address@hidden> writes:

> Hi Gaius,
> 
> With Debian Etch on an Alpha I get
> 
> # of expected passes            6336
> # of unexpected failures        30
> # of unresolved testcases       6
> 
> after applying the patches below.  Using LONGCARD instead of CARDINAL
> on 64-bit architectures seems reasonable in subaddr.mod, but there may
> be some non-obvious reason not to do that.
> 
> Michael

Hi Michael,

thanks for the patch - will apply it tomorrow!  However I've just
flushed quite a few changes to the CVS (in particular the array
handling has been altered to use gcc's ARRAY_REF tree node.
This has fixed the regression test failures in arrayhuge.mod and
arrayhuge2.mod.  On LP64 on an Athlon debian etch I now see


                === gm2 Summary ===

# of expected passes            6588

just wondering whether you get a reduced error count as well?

regards,
Gaius