[cdt-debug-dev] Re: MI questions (fwd)

["Alain Magloire" <alain@xxxxxxx>]

An exchange below that should be on this list.
Context: let Eclipse/CDT have access to raw gdb command, althought
         part of the code is not commited in CVS yet.
         The experience to "drive" the CDT/Debug/UI via the command
         is interesting, like when typing "next" and see the variables updates.


Forwarded message:
> From keiths@xxxxxxxxxx  Wed Oct 23 10:13:39 2002
> Date: Wed, 23 Oct 2002 07:24:55 -0700 (PDT)
> From: Keith Seitz <keiths@xxxxxxxxxx>
> To: Alain Magloire <alain@xxxxxxx>
> cc: cdt-debug-dev@xxxxxxxxxx
> Subject: Re: MI questions
> 
> On Wed, 23 Oct 2002, Alain Magloire wrote:
> 
> > - Can doing "b foo"  generate an event i.e. some kind of oob response?
> >   This way the UI could adjust itself and show the breakpoint.
> >   I think you mention something like that for gdb-5.4
> 
> The branch that I created a prototype of this functionality on does this, 
> but it drew objections from the community when I posted the patch for 
> approval. (Sadly, no one said anything when I initially asked about this 
> over eight months ago. Sigh.)
> 
> The objections are all based on the fact that asynchronous event 
> notifications are not associated with any specific command, i.e., the UI 
> doesn't really know which command caused what to happen. I don't think 
> this is a big deal, but there were strong objections from one group of 
> users over this, and now the whole thing appears stalled.
> 
> > - The response of "next" is not quite the same, it says:
> >   ^done,...
> >   instead of when doing -exec-next
> >   ^stopped,..
> >   I've adjust the parser, and it is quite "amusing"(meaning I like it) when
> >   you type "next" to see the rest of the UI comes alive.
> 
> Indeed it is! Definitely file a PR on this. Even on my branch I failed to 
> deal with this appropriately. It's a good thing that this is all slated 
> for gdb5.4... This one is going to take some non-trivial amount of work to 
> fix.
> 
> Keith
> 
>