Re: [Emacs-diffs] master 65797b1: Make icomplete respect `completion-ignored-extensions'

Re: [Emacs-diffs] master 65797b1: Make icomplete respect `completion-ignored-extensions'

[Stefan Monnier]

> --- a/lisp/icomplete.el
> +++ b/lisp/icomplete.el
> @@ -388,6 +388,9 @@ matches exist."
>  	(progn ;;(debug (format "Candidates=%S field=%S" candidates name))
>  	       (format " %sNo matches%s" open-bracket close-bracket))
>        (if last (setcdr last nil))
> +      (when (and minibuffer-completing-file-name
> +                 icomplete-with-completion-tables)
> +        (setq comps (completion-pcm--filename-try-filter comps)))
>        (let* ((most-try
>                (if (and base-size (> base-size 0))
>                    (completion-try-completion

Yuck.  We should really extend completion-tables yet a bit more
(probably via the completion-metadata or completion-extra-properties) so
we don't need to check minibuffer-completing-file-name and we can use
similar features in other completions.


        Stefan

Re: [Emacs-diffs] master 65797b1: Make icomplete respect `completion-ignored-extensions'

[Lars Ingebrigtsen]

Stefan Monnier <monnier@IRO.UMontreal.CA> writes:

>> --- a/lisp/icomplete.el
>> +++ b/lisp/icomplete.el
>> @@ -388,6 +388,9 @@ matches exist."
>>  	(progn ;;(debug (format "Candidates=%S field=%S" candidates name))
>>  	       (format " %sNo matches%s" open-bracket close-bracket))
>>        (if last (setcdr last nil))
>> +      (when (and minibuffer-completing-file-name
>> +                 icomplete-with-completion-tables)
>> +        (setq comps (completion-pcm--filename-try-filter comps)))
>>        (let* ((most-try
>>                (if (and base-size (> base-size 0))
>>                    (completion-try-completion
>
> Yuck.  We should really extend completion-tables yet a bit more
> (probably via the completion-metadata or completion-extra-properties) so
> we don't need to check minibuffer-completing-file-name and we can use
> similar features in other completions.

Should I revert the patch?

-- 
(domestic pets only, the antidote for overdose, milk.)
   bloggy blog: http://lars.ingebrigtsen.no

Re: [Emacs-diffs] master 65797b1: Make icomplete respect `completion-ignored-extensions'

[Stefan Monnier]

>> Yuck.  We should really extend completion-tables yet a bit more
>> (probably via the completion-metadata or completion-extra-properties) so
>> we don't need to check minibuffer-completing-file-name and we can use
>> similar features in other completions.
> Should I revert the patch?

No, that "yucky stuff" is already present at other places in
minibuffer.el.  Your patch didn't make it significantly worse.
It should makes it yet a bit more obvious that a better solution is
called for.

I didn't add a solution to this problem originally because it seemed to
be a lot of trouble, but in the mean time I added the
completion-metadata and completion-extra-properties stuff, so I think it
should be reasonably easy now to use those to implement that
new feature.


        Stefan

Re: [Emacs-diffs] master 65797b1: Make icomplete respect `completion-ignored-extensions'

[Stefan Monnier]

> No, that "yucky stuff" is already present at other places in
> minibuffer.el.  Your patch didn't make it significantly worse.
> It should makes it yet a bit more obvious that a better solution is
     ^^^^^^
     just

Sorry,

        Stefan