Re: [1/4] RFC: refactor partial symbol tables

[Tom Tromey <tromey at redhat dot com>]

>>>>> "Daniel" == Daniel Jacobowitz <dan@codesourcery.com> writes:

Daniel> Overall, I very much like this series and think it should go in.

Thanks.

Daniel> Have you considered the case of objfiles with multiple formats of
Daniel> debug info?  This used to be a very common case, where the system
Daniel> libraries had some stabs data and the user code had dwarf-2.  It's
Daniel> probably still happening on various platforms.  So it seems like
Daniel> we need to be able to query multiple symbol readers' qf functions,
Daniel> potentially.

I did consider this, and I (safely) punted.

On my branch, the only non-psymtab implementation of the quick API is in
dwarf2read.c, and that code bails out and reverts to using psymtabs if
it find that any other reader has created psymtabs for the objfile.

I'm not sure how hard it would be to do something nicer.  I didn't look
closely but I think that not all of the APIs compose nicely.

Tom