Re: calc.at workaround for current test failures

[Frank Heckenbach]

Paul Eggert wrote:

> Frank Heckenbach <address@hidden> writes:
> 
> > Paul Eggert wrote:
> > > Could you please do it?  That would save me a bit of time.
> > 
> > Patch attached....
> > I'm also doing it for abort (YY_ABORT) since programs may like to
> > provide their own version.
> 
> That's a different change, which should be addressed in a separate
> patch.  I don't offhand see the need for it, though.  Perhaps you
> can elaborate?

Some programs use things like fancy_abort() to print more readable
messages in case of fatal failure. (I understand that such functions
must really abort, not return after printing an error, but in the
cases I know, that's so.)

> > Besides YYSTACK_ALLOC and YYSTACK_FREE (copied from yacc.c), glr.c
> > also needs permant allocation. I used the macro names YY_MALLOC,
> > YY_REALLOC and YY_FREE, is this ok?
> 
> I'd prefer a 'YY' prefix instead of a 'YY_' prefix, since that's the
> convention.  (YY_ABORT would be a special case, though, since YYABORT
> is taken.)
> 
> Also, yacc.c should be modified similarly,

Oh yes. (I had missed it because my system has alloca, so I didn't
get that warning, but you're right, of course.)

> and glr.c can remove a few
> casts.  I don't think we need the awkward and portability-challenged
> YYSTACK_* macros in glr.c, as the only place they can be used is in a
> rarely-used location, so the efficiency motivation for them isn't
> there.

Agreed. It was probably overkill.

> So, how about the following change instead?  (I'm trying to illustrate
> the preferred form for discussing patches: a single subject, or at
> least closely related subjects, with a ChangeLog entry before the
> patch.)

I hope the following patch for YY_ABORT meets the requirements. (And
since my assignment has now been processed by the FSF, there are
hopefully no more obstacles to including it.)

Frank

-- 
Frank Heckenbach, address@hidden
http://fjf.gnu.de/
GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)

glr-abort.diff
Description: Binary data