Re: [RFA 1/2] Linespec rewrite (update 2)

[Tom Tromey <tromey@redhat.com>]

>>>>> "Keith" == Keith Seitz <keiths@redhat.com> writes:

Keith> I don't know. The whole comma thing is undocumented. The test suite
Keith> does contain list ranges. That's how I originally discovered
Keith> this. I've removed the list mode restriction, though, and it doesn't
Keith> affect test results at all.

You can write a test case using python that calls gdb.decode_line and
examines the remainder of the line.

Comma-termination isn't documented but I think it has to be preserved
anyway.

Keith> Yes, we can end up with a canonical form like "function:+5" or
Keith> "file:+5". The former is permitted (per recent maintainer request)
Keith> because we currently ignore the offset. [It is unprocessed in
Keith> convert_linespec_to_sals.] I'm not a fan of this,

What is the rationale for having a linespec where parts are ignored?
I couldn't think of a use for it.  And, if current cvs rejects it, then
it seems like it is interfering with a useful future feature as well.

Tom> Why are minsyms sorted by pspace in one branch but not another?

Keith> No real good reason, other than that is the way it is done today. I
Keith> tried to keep the codepaths as similar as possible. I've merged the
Keith> two branches together. No need for minsyms to be singled out like
Keith> this.

Thanks.  Keeping things the same is sufficient rationale, but now that
you've merged it, that is fine too.

Tom