Re: mi_load_progress question

[Daniel Jacobowitz <drow at false dot org>]

On Wed, Jul 12, 2006 at 09:32:34PM +1200, Nick Roberts wrote:
> Vladimir Prus writes:
>  > 
>  > Hi,
>  > I see some strange logic in the mi_load_progress function (file mi/mi-main.c). 
>  > That function is responsible for printing progress report when downloading 
>  > program to target, and the code in question is:
>  > 
>  >  if (current_interp_named_p (INTERP_MI))
>  >     uiout = mi_out_new (2);
>  >   else if (current_interp_named_p (INTERP_MI1))
>  >     uiout = mi_out_new (1);
>  >   else
>  >     return;
>  > 
>  > When I run gdb with "--i=mi2", this code exists with "return", producing to 
>  > progress information. Is this desired behaviour?

I am pretty sure that there was a bug report about this before, but I
can't find it now.  Ah, here:

http://sourceware.org/ml/gdb-patches/2005-11/msg00295.html

No one else commented at the time and I opted to stick with what we
already had.  Do you think we should just turn the progress reporting
back on for -i=mi2?  How about you, Nick?

> Looking through the change history, other values didn't exist when the original
> code was written.  So I guess it's not desired behaviour.
> 
>  if (current_interp_named_p (INTERP_MI1))
>    uiout = mi_out_new (1);
>  else if (current_interp_named_p (INTERP_MI)
> 	  || current_interp_named_p (INTERP_MI2))
>    uiout = mi_out_new (2);
>  else if (current_interp_named_p (INTERP_MI3))
>    uiout = mi_out_new (3);
>  else
>    return;
> 
> would work (until new MI levels are introduced!).  It would be best to generalise
> this and similar code to make it future proof.

Or come up with some other way to do this that avoids the need for the
hack.  The problem was that we were calling the MI load progress hook
through the wrong interpreter.  We need to figure out how to do this
right.

First question: Should typing "load" at the MI prompt issue +download
updates?  Currently it does.  It also issues the CLI's updates.

Second question: Should typing "-interpreter-exec console load" issue
+download updates?  Currently it does the same as "load" and
that's the right thing to do so this is the same as the first question.

Jim, maybe you've got an opinion on this?  Apple seems to make more use
of the CLI support than anyone else :-)

-- 
Daniel Jacobowitz
CodeSourcery