> Date:          Tue, 8 Apr 1997 21:05:36 -0400
> From:          Bill Dubuque <martigny.ai.mit.edu at wgd>
> To:            <truesoft.com at dtenny>
> Cc:            <Franz.COM, at editi> <cs.berkeley.edu, at allegro-cl> <games.ultranet.com, at tenny>
>                <martigny.ai.mit.edu at wgd>
> Subject:       Re: EDITI 3.0 Feedback: enhancement request

> 
> Dave Tenny <truesoft.com at dtenny> wrote to <Franz.com at Editi> on 8 Apr 1997:
> : 
> : ... Usually when I want to TRACE or BREAKPOINT a function,
> : I'm already sitting with the point somewhere in the definition
> : of the function I want to trace, as opposed to the call
> : site for that function. ...
> 
> I agree this would be a useful capability (though it would require
> restructuring the prefix arg structure for all of the Tools commands
> in Editi 3.0).
> 
> Perhaps it would make sense to have the Tools commands (trace,
> breakpoint, profile ...) select the sexp in the same manner as the 
> evaluation commands, namely 
> 
>    arg = 0,  the symbol at point, e.g. C-0 C-T
>    arg = 1,  the sexp containing point (or just before/after point)
>    arg = 4,  the top-level form containing point (same as 'arg none')
> 
> Then the functionality you request would be simply  C-4 C-T,
> or more simply  C-T.

I like it, but it would be a shame to lose all the existing
work.  I'm less picky about key bindings, I can always change
those.  But I think both forms of interaction are truly
useful, it simply depends on the mode of debugging you're
employing at a given moment.

Mine is *usually* one where I want to trace the function
whose definition I'm looking at.  But even for me, and I'm
sure for others, I sometimes am looking at a flow of function
calls and want to trace the calls based on call site.