Re: MCL 3.0 wishes

[eliot@cs.umass.edu (CHRISTOPHER ELIOT)]

In article <9310202246.AA03359@relay1.UU.NET>, dhm%proact@uunet.UU.NET ("Don Mitchell") writes...
>Subject:   RE>MCL 3.0 wishes
>I find the debugger very effective especially with its easy interface to the
>inspector.
>My symbols are never gibberish because I always use [fasl-]save-....  The
>disassembly listing is easy to follow (double-click on the function in
>the backtrace)

I tried this 60 seconds ago and it does not bring up a *disassembly* 
listing - it brings up the inspector on the function.  This just
shows the name and arglist in a window, which is not very helpful to
me.

>and figure out where you are in your
>source because it has all the function and variable names.  Double click on
>any function symbol or local variable, and you're inspecting it.

That's what it seems to do.  But again, how do you get a dissasembly
of a function?  For large functions it is often very difficult to
find out where you are and the Lisp Machine debugger would just
hand you the current location inside the function right there
on a silver platter.  MCL does not.

Don't get me wrong, I think that MCL is a great Lisp environment,
but the debugging facilities are just not as helpful as the Lisp
machine.  I can live with it, I can understand it, but I've seen
a better debugger.  At best the MCL debugger will get you to
the offending function but it really does not give you any
information about where within that function to look.

In *most* situations there will happen to be additional *clues*
as to where you are, and I never write huge monololothic functions,
but the computer has a return address or a program counter sitting
right there are I wish it would just tell me what it says so
I don't have to infer the information from clues.

In general MCL is great.  Almost all of its weaknesses are a direct
reflection of weaknesses of the Macintosh "operating system".
But the debugger needs work.  It is at the top of my list of problems
with MCL.


>I used Symbolics and TI lisp machines for 10 years and found the transition
>to MCL easier than the transition to any of the Unix lisps.  

Same here, except I have not yet been forced to deal with Unix.

Chris Eliot