This is the mail archive of the gdb-patches@sourceware.org mailing list for the GDB project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [RFC] [1/2] First cut at multi-executable support.


>>>>> "Pedro" == Pedro Alves <pedro@codesourcery.com> writes:

Pedro> We could split an objfile in two, with the shared parts not having a
Pedro> reference to a symbol space. [...]

Pedro> At a high distance, this sounds feasible to me, but, looking at the
Pedro> code, this looks like a huge effort, and I'm sure I'm missing a lot
Pedro> of complications.

Yeah.  I just find it helpful to have an overview of the wish-list.

Tom> This seems like another possible performance problem; lazily reading
Tom> it would be friendlier.  In a scenario like the "make" case, I would
Tom> assume that most inferiors will not actually require any user
Tom> attention, and time and memory spent on their debuginfo is just
Tom> wasted.

Pedro> Yeah, I know you have patches for this.  ;-)

Hah, yeah, I was kind of plugging my patch.  I'm hoping that we can
actually push this lazy reading idea much further -- with my patch,
the pause when gdb actually goes to read the debuginfo can be rather
noticeable on occasion.  But, this is vague dreaming, I don't really
know how to implement full laziness yet.

Tom> "Program exited normally." could also use some love... at least some
Tom> info about the program, and maybe removing the excess newlines?

Pedro> For my own testing, I've tweaked those messages to include the
Pedro> process id.  It's just mad otherwise.  However, it has been
Pedro> one of my goals to not change much how the single-inferior case
Pedro> works/outputs as a first incremental step.  I'm sure there will
Pedro> be different opinions on to what GDB should output, so this avoids
Pedro> such discussions for now :-).

Practical plan :-)

Tom> +  ui_out_table_header (uiout, 1, ui_left, "current", "");

Tom> For some reason I did not think of leaving the header empty when I did
Tom> this for "info inferiors".  I like this better... since you're also
Tom> touching print_inferior, how about just making that change?

Pedro> But I had already.  :-)  Or, do you mean to split that change out
Pedro> of the patch?  I can do that.  I think that would be good idea.

I thought your patch had this change in one place but not "info
inferiors".  Maybe I missed it.  Anyway, don't worry about it, I'll
mark this note and go back and check once everything goes in.

Pedro> I'm now working on cleaning up a bit the patch and fixing the
Pedro> random crashes I mentioned, and I'll be posting an updated
Pedro> patch soon.

I'm looking forward to it.

Tom


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]