Re: [patch] Bug #5001 - ctime() vs _stp_ctime() duplication

[Jim Keniston <jkenisto at us dot ibm dot com>]

On Wed, 2008-05-21 at 14:57 +0200, Mark Wielaard wrote:
> Hi,
> 
> On Wed, 2008-05-21 at 14:01 +0200, Mark Wielaard wrote:
> > I am not sure how I should fix this. Is the idea that the arguments are
> > now always referred to by number?

There's been no consensus or fiat to convert tapsets from named args to
numbered args so they can be used in dwarfless probing.  But there's
been some background work on doing just that for the syscalls tapset --
see nd_syscalls.stp -- and I have a CS student who's willing to continue
that conversion once spring term ends.

In this case, the fault may be with the nodw*.stp tests.  There's no
particular reason nodwf02.stp and nodwf05.stp have to probe exactly that
set of system calls.  If your changes to syscalls*.stp make sense
otherwise, then we should just tweak the list in nodwf0{2,5}.stp.

Sorry, I haven't been following the 5001 thread.  Could you send me
before and after tapsets so I can figure out what went wrong wrt
dwarfless?

Thanks.
Jim

> 
> So, the patch makes it work also "dwarfless". If this doesn't show any
> regressions on a full make installcheck I'll commit and push it. But I
> don't really understand why the previous version worked.  The only thing
> I changed in my original patch is how the arguments were used, now how
> they were fetched. If anybody has an explanation that would be
> appreciated.
> 
> Thanks,
> 
> Mark