bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

[Jose Marino]

I'm using idlwave mode with emacs 24.2 and filename completion in the 
shell is not working. Typing TAB produces nothing: no errors and no 
completions.

The problem seems to be fixed by this patch:
--- idlw-shell.el.orig	2012-09-11 13:35:35.549891109 -0500
+++ idlw-shell.el	2012-09-11 13:36:37.901886841 -0500
@@ -2187,7 +2187,7 @@
    ;; CWD might have changed, resync, to set default directory
    (idlwave-shell-resync-dirs)
    (let ((comint-file-name-chars idlwave-shell-file-name-chars))
-    (comint-filename-completion)))
+    (comint-dynamic-complete-filename)))

  (defun idlwave-shell-executive-command ()
    "Return the name of the current executive command, if any."


Thanks,
Jose

bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

[Glenn Morris]

Version: 24.3

Jose Marino wrote:

> -    (comint-filename-completion)))
> +    (comint-dynamic-complete-filename)))

Thanks; applied to emacs-24 branch.


The obsolescence message for comint-dynamic-complete-as-filename seems
misleading to me:

   This function is obsolete since 24.1;
   use `comint-filename-completion' instead.

Surely it would be better to say `use comint-dynamic-complete-filename
instead', rather than just saying "See `comint-dynamic-complete-filename'"
in the doc.

bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

[Stefan Monnier]

>> -    (comint-filename-completion)))
>> +    (comint-dynamic-complete-filename)))

> The obsolescence message for comint-dynamic-complete-as-filename seems
> misleading to me:

>    This function is obsolete since 24.1;
>    use `comint-filename-completion' instead.

> Surely it would be better to say `use comint-dynamic-complete-filename
> instead',

Actually, that depends on what comint-dynamic-complete-as-filename was
used for: if it was used as a *command*, then
comint-dynamic-complete-filename is a good replacement.  But if it was
used within something like comint-dynamic-complete-functions, then the
better replacement is comint-filename-completion.

IOW, in the case of idlw-shell.el, the better replacement is indeed
comint-filename-completion, tho it requires changes in
idlwave-shell-complete (basically making it an alias to
completion-at-point and turning its body into a list of functions to put
on completion-at-point-functions).


        Stefan

bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

[Glenn Morris]

Stefan Monnier wrote:

> Actually, that depends on what comint-dynamic-complete-as-filename was
> used for: if it was used as a *command*, then
> comint-dynamic-complete-filename is a good replacement.  But if it was
> used within something like comint-dynamic-complete-functions, then the
> better replacement is comint-filename-completion.

Could you make the obsolescence message less definitive-sounding then?

> IOW, in the case of idlw-shell.el, the better replacement is indeed
> comint-filename-completion, tho it requires changes in
> idlwave-shell-complete (basically making it an alias to
> completion-at-point and turning its body into a list of functions to put
> on completion-at-point-functions).

Hopefully you feel like doing that (or reopening this report).

bug#12418: emacs 24.2; idlw-shell.el: filename completion in IDL shell is not working

[Stefan Monnier]

>> Actually, that depends on what comint-dynamic-complete-as-filename was
>> used for: if it was used as a *command*, then
>> comint-dynamic-complete-filename is a good replacement.  But if it was
>> used within something like comint-dynamic-complete-functions, then the
>> better replacement is comint-filename-completion.
> Could you make the obsolescence message less definitive-sounding then?

I don't think it's worth the trouble: the use of
comint-dynamic-complete-filename is very rare.

>> IOW, in the case of idlw-shell.el, the better replacement is indeed
>> comint-filename-completion, tho it requires changes in
>> idlwave-shell-complete (basically making it an alias to
>> completion-at-point and turning its body into a list of functions to put
>> on completion-at-point-functions).
> Hopefully you feel like doing that (or reopening this report).

I don't really feel like it, but I will probably be the first to get to
it, regardless of how slowly I do it :-(


        Stefan