Re: are there outstanding patches on the compiler side for gdb namespace work?

[David Carlton <carlton at math dot stanford dot edu>]

On Fri, 28 Feb 2003 11:28:53 -0600, Benjamin Kosnik <bkoz at redhat dot com> said:

> What needs to happen compiler-side for g++ to give gdb the proper info?

GCC needs to emit DW_TAG_namespace, DW_TAG_imported_declaration, and
DW_TAG_imported_module info; no released version of GCC does.  Which
is GDB's fault, not GCC's; but it's safe now that GDB 5.3 has been
released, since GDB 5.3 at least won't be confused by their presence.
I'm cautiously optimistic that the next version of GDB will do
something useful with DW_TAG_namespace (and actually be able to
sometimes figure out where it should be present even if it isn't
there): I've started submitting RFAs to get such code added to GDB's
mainline.

The last I heard, it should be easy enough for GCC to emit
DW_TAG_namespace, but that it's hard to handle the other two tags
cleanly with the way that the front end and back end currently divide
their work: the back end doesn't quite have enough information.  This
is all second-hand from Daniel Berlin; he's probably a better person
to ask about this.

I don't know what debug info the current GCC 3.3 and mainline branches
produce: I don't read gcc-patches, and I only just downloaded the
mainline (and almost immediately ran into a show-stopping exception
handling problem (PR c++/9756, I guess?), but never mind that).

Incidentally, if anybody out there is actually using GDB's
carlton_dictionary-branch (sorry about the inconsistent
punctuation...) and has bug reports for it or suggestions about how it
could be improved, I'm all ears.  Or, even if you aren't using it but
have ideas about what you'd like GDB to do with namespaces that it
currently isn't, that would be lovely as well.

David Carlton
carlton at math dot stanford dot edu