RE: Regression for gdb.pascal/* [Re: [RFA 4/4] Constify parse_linesepc]

["Pierre Muller" <pierre.muller@ics-cnrs.unistra.fr>]

> -----Message d'origine-----
> De : gdb-patches-owner@sourceware.org [mailto:gdb-patches-
> owner@sourceware.org] De la part de Tom Tromey
> Envoyé : mardi 29 octobre 2013 17:39
> À : Jan Kratochvil
> Cc : Pierre Muller; Keith Seitz; gdb-patches@sourceware.org ml; Sergio
> Durigan Junior
> Objet : Re: Regression for gdb.pascal/* [Re: [RFA 4/4] Constify
> parse_linesepc]
> 
> >>>>> "Jan" == Jan Kratochvil <jan.kratochvil@redhat.com> writes:
> 
> Jan> Maybe we could juse use
> Jan> 	[pascal patch] Use case_sensitive_off [Re: Regression for
> Jan> gdb.pascal/* [Re: [RFA 4/4] Constify parse_linesepc]]
> Jan> 	https://sourceware.org/ml/gdb-patches/2013-10/msg00581.html
> Jan> and forget about all the case changes then.
> 
> I think this would be the best approach.
> 
> However, I think it would be good for Pierre to weigh in.
> Pierre, you replied on this thread but I don't think you addressed
> Jan's
> patch, which seems like the best way forward.

   I am sorry, but I have not enough time lately to 
look more deeply into this problem...

   One thing I do remember was about the fact
that some of the internal procedures/functions used by Free Pascal
compiler were written with lowercased assembly labels
to avoid conflicts with usual procedures or functions

   Is Keith patch already in latest git?
OK, it seems Keith last patch is not in,
while it does indeed fix otherwise illegitimate stack access
(alloca memory read after exiting function using it...)
   So Keith patch should go in.

  Concerning Jan's patch, I am more hesitant because
the fact to change the default behavior of 
pascal parser to  case-insensitive does have noticeable behavior changes
for end-users...

  I did not see any difference in the testsuite between
Keith and Jan's patch, as Keith patch alone already get's us back to only
one failure in the test "p 0x1.1"
I must admit that Free pascal never accepted to parse 
hexadecimal reals, is this accepted in any other pascal compiler ?
According to 'git blame' this test was added by Jan on 2010-09-06...
  Second related question is: there is an xfail in the test,
but I stil get FAIL here... Shouldn't I get XFAIL?


> If Pierre can't respond soon-ish, I think we should go ahead in the
> interests of forward motion and regression-fixing; and deal with
> whatever fallout arises when he returns.

As I said, let's commit Keith part
and wait a little bit on the case insensitive change...
I would also like to fix the completion bug in between
to check if Jan's patch changes anything in that part..

Pierre
as pascal language maintainer