Re: update on speedup project

[Roland McGrath <roland at redhat dot com>]

> 1. A by-name search reaches this objfile.
> 
> 2. A by-address search needs partial symbols for this objfile.
>    This seems needed for backtraces.

Right.  I just wasn't sure if you had in mind some more complicated
definition of when a by-name search does or doesn't reach this objfile.
If it's literally no change to the search algorithm, that's fine.
My concern was to keep an eye out for any subtleties such as "find the
first match of what's already interned" code that works now because
everything is interned first, vs cases where it could appear to work
because enough was interned to get some answer and mask the info it would
get by interning lazy files.


Thanks,
Roland