[RFA, doc RFA] Avoid calling gdb_realpath if basenames are different

[Doug Evans <dje at google dot com>]

On Sat, Nov 5, 2011 at 11:30 PM, Doug Evans <dje@google.com> wrote:
> Hi.
> This patch has been brought up before (by others).
> E.g., http://sourceware.org/ml/gdb-patches/2010-04/msg00466.html
> I'm hoping we can get this in now.
> We're paying a real and significant cost for what is mostly a
> theoretical concern.
> [E.g., How often is one source file referred to by the user using a basename
> that is different than what's recorded in the debug info?]
>
> If people are concerned about breaking someone's usage,
> we could default basenames-may-differ to true in 7.4,
> with a warning that it will be set to false in 7.5 (or some such).
> [We could leave the default set to true, especially if someone knew
> of at least some minimally common usage this would break.
> I'd hate to otherwise penalize the vast majority of users if not.]

Hi.
Ok to check in?

Note: I set the default to be the common case (speed up gdb by
assuming basenames never differ).
Let me know if you want the default changed.

Tom: I'll look at the bugs you mentioned separately.

2011-11-10  Doug Evans  <dje@google.com>

        * NEWS: Mention new parameter basenames-may-differ.
        * dwarf2read.c (dw2_lookup_symtab): Avoid calling gdb_realpath if
        ! basenames_may_differ.
        * psymtab.c (lookup_partial_symtab): Ditto.
        * symtab.c (lookup_symtab): Ditto.
        (basenames_may_differ): New global.
        (_initialize_symtab): New parameter basenames-may-differ.
        * symtab.h (basenames_may_differ): Declare.

        doc/
        * gdb.texinfo (Files): Document basenames-may-differ.

Attachment: gdb-111110-basenames-may-differ-2.patch.txt
Description: Text document