Re: [PATCHv3] gdb: remove attempted type de-duplication when building gdb-index

[Tom Tromey <tom@tromey.com>]

>> This also seems like something that could be fixed.  For instance
>> completion could be done from the index as first pass.

Andrew> To be clear, this isn't just a completion issue.  The example in PR
Andrew> gold/15646 is for 'break namespace_foo::misspelled', where this is also
Andrew> going to trigger expansion of many CUs.

Andrew> But maybe there's information that we could track better that would
Andrew> allow us to only expand a single CU in some cases?

For completion I had some thoughts, namely that the only time we need to
expand a CU for completion is when completing a C++ function, because I
think that's the only case where an index entry and a full symbol should
have different names.

For the "break prefix::fn" case, maybe if linespec sees that "prefix" is
a namespace it could stop the outer search.

It might still be an issue if "prefix" is a class since I think (1)
linespec may try to set breakpoints in superclass methods (not sure) and
(2) different CUs may have different views on a class, like different
template method instantiations.  However at least (2) is a case of
getting what you asked for.

Tom