[Axiom-developer] Re: [ANN] yet a better version of the axiom mode for emacs.

[Martin Rubey]

Jay Belanger <address@hidden> writes:

> Let me echo the praise you're getting; this is very nice.
> 
> >   In fact, all of this waiting-for-output stuff should be rewritten, but I
> >   don't understand it enough.
> 
> It's a pain; is it causing problems?  

Mainly in so far as I do not understand really what it is doing.  And the
initialization code is a mess.  For example, why are there two functions hooked
to comint-output-filter-functions, namely:

  axiom-output-filter (str)
  make-output (str)

(The latter had a different name in an earlier version.)

Why is (sit-for 0 axiom-after-output-wait) necessary?

Is (setq comint-prompt-regexp axiom-prompt) still useful?  I am now marking
prompts as 'field, since (beginning-of-line) didn't work otherwise.  (sometimes
it jumped over the prompt, sometimes it didn't).

> I think it should be kept in, but for the time being (until someone needs it
> for EAxiom or something) it could be made to do nothing by commenting out
> `axiom-wait-for-output' and replace it by a function with the same name which
> does nothing.

Well, that doesn't work either, see below.

> Or even make it configurable:
> make a variable `axiom-output-wait' or something, and 
>   (defun axiom-wait-for-output ()
>      (if axiom-output-wait
>           previous definition
>        nil))
> 
> 
> > * For me, the following is *very* severe. Try:
> >
> >     for i in 1..10 repeat ([j for j in 1..2000]; output "hi")
> >
> >   in a usual shell buffer and in axiom mode.  Very unfortunately, the axiom
> >   mode currently accumulates all output and then displays it at once.  
> > Since I
> >   use such constructs often to be able to check how far a computation got
> >   already, it makes the mode unusable for me.
> 
> This has nothing to do with waiting for output, or even Emacs.  Axiom
> will do that when called with the -noclef option; I have no idea why.

Well, at least in wh-sandbox, this is *not* the case.  And if I remove all
these accept-process-output stuff, it works, too.  However, I have then some
problems deciding when to decide that a prompt arrived, since it seems that
emacs continues execution of my elisp code, even if output is still arriving.

Many many thanks,


Martin