[drow-cplus-branch] lookup_symbol_aux refactoring

[drow-cplus-branch] lookup_symbol_aux refactoring

[David Carlton]

Is it okay if I commit the lookup_symbol_aux refactoring from 
<http://sources.redhat.com/ml/gdb-patches/2002-10/msg00080.html> to
drow-cplus-branch?  I've got some patches to symbol table lookup to
handle namespace stuff that are almost ready (I just want to add some
testcases), and they'll be a lot cleaner if I get this part of them
out of the way first.

David Carlton
carlton@math.stanford.edu

Re: [drow-cplus-branch] lookup_symbol_aux refactoring

[Daniel Jacobowitz]

On Fri, Oct 11, 2002 at 09:07:51AM -0700, David Carlton wrote:
> Is it okay if I commit the lookup_symbol_aux refactoring from 
> <http://sources.redhat.com/ml/gdb-patches/2002-10/msg00080.html> to
> drow-cplus-branch?  I've got some patches to symbol table lookup to
> handle namespace stuff that are almost ready (I just want to add some
> testcases), and they'll be a lot cleaner if I get this part of them
> out of the way first.

Hmm... yes, this will be useful for me, too.  The only bit I'm nervous
about is the changed search order, but we can update this to match
whatever goes in mainline when your patch is approved.  Please do.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: [drow-cplus-branch] lookup_symbol_aux refactoring

[David Carlton]

On Fri, 11 Oct 2002 12:16:43 -0400, Daniel Jacobowitz <drow@mvista.com> said:
> On Fri, Oct 11, 2002 at 09:07:51AM -0700, David Carlton wrote:

>> Is it okay if I commit the lookup_symbol_aux refactoring from 
>> <http://sources.redhat.com/ml/gdb-patches/2002-10/msg00080.html> to
>> drow-cplus-branch?  I've got some patches to symbol table lookup to
>> handle namespace stuff that are almost ready (I just want to add some
>> testcases), and they'll be a lot cleaner if I get this part of them
>> out of the way first.

> Hmm... yes, this will be useful for me, too.  The only bit I'm nervous
> about is the changed search order, but we can update this to match
> whatever goes in mainline when your patch is approved.  Please do.

Done.  Yeah, I'm not convinced that the changed search order is
ultimately the correct thing to do either.  What is important for me
right now is to have an easy way to look up a symbol in the global
environment.  Unfortunately, the way minimal symbol lookup is
currently handled screws that up, because it will use static symbols as
a fallback measure.

So ultimately that will have to get sorted out, and we can go with
whatever solution is adopted on the mainline.  And I honestly have no
idea what that solution will be: it seems to me that trying to use all
three of symtabs, psymtabs, and minimal symbols is probably redundant,
but maybe I'm missing something and even if they are I don't have a
solid feeling for the best way to remove some of the redundancy.

For now, I'm adopting the tactic of making sure that things work well
if there's proper debugging info but also making the code modular
enough that that aspect of the search strategy will be easy to
change.

David Carlton
carlton@math.stanford.edu