Re: [Reduce-algebra-developers] SIGINT handling of CSL Reduce inside Emacs Reduce-IDE and TeXmacs interfaces

[Andrey I. <and...@ma...>]

Hi, Francis,

First of all thank you very much for REDUCE-IDE! I'm very happy that 
Reduce has its own Emacs mode and interface to the interpreter.

My platform is Linux. Currently I am using Debian 12 (Emacs 28.2) and 
GNU Guix (Emacs 29.1).

Andrey.

Francis Wright wrote:
> Hi, Andrey
> 
> Thanks for your email. I'll consider your suggestions for the next release of REDUCE-IDE. What platform(s) are you using? I think I added the code to remove ansi-color-process-output a very long time ago and the reason probably related to an old version of MS Windows. I can probably now remove it, as least on some platforms.
> 
> Best wishes,
> Francis
> 
> ________________________________
> From: Andrey Ignatenko via Reduce-algebra-developers <red...@li...>
> Sent: 25 January 2024 2:56 PM
> To: red...@li... <red...@li...>
> Subject: [Reduce-algebra-developers] SIGINT handling of CSL Reduce inside Emacs Reduce-IDE and TeXmacs interfaces
> 
> Hello all,
> 
> I hope the following may be of interest both to users and developers.
> 
> 1. CSL version of Reduce, when used inside Emacs via Reduce-IDE
> interface simply terminates on SIGINT (interrupt) signal generated,
> e.g., by C-c C-c keystroke. Very similar behaviour is observed for CSL
> Reduce in TeXmacs session when the calculation is interrupted by
> pressing to the 'Stop' button.
> 
> I find this behaviour of redcsl inside above-mentioned very useful
> Reduce interfaces quite inconvenient. Below I suggest a couple of simple
> hacks that allow to get rid of this for Emacs Reduce-IDE.
> 
> 2. First of all, this behaviour is not specific to Reduce-IDE. The same
> is observed when 'redcsl -w' is run inside Emacs shell (M-x shell)
> buffer. Moreover, it is not specific to Emacs at all since it manifests
> themselves when Reduce is run inside ordinary terminal via command 'env
> TERM=dumb redcsl -w' which sets environment variable TERM to 'dumb'.
> Inside Emacs shell the variable TERM is also set to 'dumb' by default.
> 
> 3. The idea of the hack is to trick redcsl to think that it was run
> inside more sophisticated terminal when a 'dumb' terminal. You can
> archive this by setting 'reduce-run-commands Reduce-IDE variable to
> (("CSL" . "env TERM=<termname> redcsl --nogui") ("PSL" . "redpsl")) with
> <termname> replaced by one of the terminals known to your system. For
> example, you can try "eterm-color", the default value of TERM variable
> inside Emacs terminal (M-x term). With this modification Reduce-IDE (M-x
> run-reduce) now reacts properly to C-c C-c interrupts.
> 
> However, Reduce-IDE does not handle terminal color codes now emitted by
> redcsl. There are two solution to this problem. First, one can use
> <termname> corresponding to terminal that does not support color codes,
> e.g. "vt100" or "vremote". Second, one can modify source code of
> Reduce-IDE by commenting out the following two lines
> 
>    (remove-hook (make-local-variable 'comint-output-filter-functions)
>                 'ansi-color-process-output t) ; remove locally!
> 
> in "reduce-run.el" file, thus allowing Reduce-ide to process color
> codes. Above these two lines there is the following comment
> 
>     ;; ansi-color-process-output causes an error when CSL is terminated
>     ;; and is probably irrelevant anyway, so ...
> 
> However, I do not observe any errors with CSL termination on Emacs 28.2.
> Maybe the time has come to remove these lines from Reduce-IDE package?
> 
> Regards,
> Andrey Ignatenko
> 
> 
> _______________________________________________
> Reduce-algebra-developers mailing list
> Red...@li...
> https://lists.sourceforge.net/lists/listinfo/reduce-algebra-developers
>