Re: [PATCH] Verify COFF symbol stringtab offset

[Tom Tromey <tom@tromey.com>]

>>>>> "Keith" == Keith Seitz via Gdb-patches <gdb-patches@sourceware.org> writes:

>> If it is really that dead, then -- like stabs -- I'd like to remove
>> it.  These dead formats just make it harder to improve gdb while not
>> providing value to any user.  Actually with the advent of fuzzers,
>> this code now has negative value, because it wastes time coming up
>> with patches to fix bugs in code no one uses.

Keith> I was tempted to suggest either removing it or at least adding a
Keith> --disable-reader-coff option to disable the code, but I could not
Keith> convince myself that there wouldn't be relatively widespread
Keith> disagreement.

If we think the format is truly obsolete, then IMO it is better to just
delete it.

I never got back to replying about the stabs situation, but after Pedro
weighed in, I think we should move ahead with that.  However, we're
close-ish to GDB 14 now, so I plan to wait until after that branches.

The main problem for other formats is even knowing if they are in use.
Searching for buildsym-legacy, I see coffread (I guess COFF had some
kind of debuginfo?), dbxread (a.out but maybe stabs-like?), mdebugread
(ECOFF), xcoffread (yet another COFF).

My guess is that maybe only xcoffread could even be remotely relevant
but I don't really know how to tell.

Tom