This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: Resend: [RFC] p-exp.y enhancements.
- From: Pierre Muller <muller at cerbere dot u-strasbg dot fr>
- To: Michael Snyder <msnyder at redhat dot com>
- Cc: gdb-patches at sources dot redhat dot com
- Date: Fri, 03 May 2002 09:42:10 +0200
- Subject: Re: Resend: [RFC] p-exp.y enhancements.
- References: <4.2.0.58.20020502144659.02489e38@ics.u-strasbg.fr>
At 19:53 02/05/2002 , Michael Snyder a écrit:
>Pierre Muller wrote:
> >
> > (Second posting with the patch attached,
> > this way you should have less problems to
> > apply the patch).
> >
> > This patch has several benefits:
> >
> > 1) it adds support for S[exp] if S is a pascal string.
> > I must admit that this only works for Free Pascal Compiler
> > for now, because GPC uses BITSTRING type
> > for the array containg the chars which confuses GDB.
> > I informed the GPC developper team, but their answer
> > seems to be that they didn't really directly change any code
> > regarding debug information output for pascal...
>
>What will happen if a GPC user tries to do that?
Well, before it gave an error that there is no [] operator,
after it will give a different error.
(
> > 2) it adds support for case insensitive searches of record
> > (pascal struct) fields.
>
>This seems like a good thing...
>
>
> > 3) It prepares to a more simple RFC
> > about the enhancement to the completer for
> > structure field that I sent some time ago. See
> > http://sources.redhat.com/ml/gdb-patches/2002-04/msg00774.html
>
>This seems like a good thing too, but is pretty
>unrelated to the two above. Would you mind breaking
>this patch down into functional units? Since I'm a
>little rusty in both Pascal and Yacc, that will make
>it considerably easier to review.
No, it is clearly related because, in order to get
a way to get the completer to work of struct fields,
we need of course to know the current_type
of a part of an expression, which is the core of this patch!
> > The way current_type is treated in my patch is
> > not completely correct, in the sense that
> > current_type is often non NULL when it should be NULL
> > (like for instance after an operator like '+' or '*',
> > but this has no influence on its functionallity, so I did not
> > add all the rules that would reset current_type
> > (which would have made the patch much bigger).
> >
> > If I get no observations about this patch, I will commit it
> > in the next days.
> >
> > 2002-05-02 Pierre Muller <muller@ics.u-strasbg.fr>
> >
> > * p-exp.y (current_type): New static variable.
> > Carries the type of the expression at the position that is parsed.
> > (push_current_type, pop_current_type): Two new functions. Used
> > to store/restore current_type in expression on specific tokens.
> > (search_filed): New static variable. Set to one after parsing a point as
> > at that point only a FIELDNAME token should be searched.
> > (FIELDNAME): New token. After a point only a token belonging to
> > current_type type definition is allowed.
> > (all over token rules): reset and change current_type according
> > to rules.
> > (exp '[' rule): insert implicit array index field if exp is a pascal string type.
Pierre Muller
Institut Charles Sadron
6,rue Boussingault
F 67083 STRASBOURG CEDEX (France)
mailto:muller@ics.u-strasbg.fr
Phone : (33)-3-88-41-40-07 Fax : (33)-3-88-41-40-99