Re: master 7fc0292: Allow completion styles to adjust completion metadata

["João Távora" <joaotavora@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 1983 bytes --]

You're right, this version didn't work for many reasons. Once we get
cl-defgeneric to work, we can fix it to work by return value, as you
suggest...

João

On Sat, Oct 26, 2019, 17:12 Stefan Monnier <monnier@iro.umontreal.ca> wrote:

> > +(cl-defgeneric completion-adjust-metadata-for-style (style metadata)
> > +  "Adjust METADATA of current completion according to STYLE."
> > +  (:method (_style _metadata) nil) ; nop by default
>
> Hmm... if "nop" is to return nil, it means it works by side-effect, but
> if the metadata is `nil` to start with, there's no side-effect that can
> turn it into a non-nil value, right?
>
> So I think it should work by returning a new metadata and hence "nop"
> should be to return `metadata` unmodified.
>
> > +     (let ((alist (cdr metadata)))
> > +       (setf (alist-get 'display-sort-function alist)
> > +             (compose-flex-sort-fn (alist-get 'display-sort-function
> alist)))
> > +       (setf (alist-get 'cycle-sort-function alist)
> > +             (compose-flex-sort-fn (alist-get 'cycle-sort-function
> alist)))
>
> This works by side-effecting the alist, but those entries may (and
> usually) come from immediate constants in the code, so this is
> like "self-modifying code" :-(
>
> > +       metadata))))
>
> I see here you do return the metadata, good.
>
> > +        (result-and-style
> > +         (completion--some
> > +          (lambda (style)
> > +            (let ((probe (funcall (nth n (assq style
> > +                                               completion-styles-alist))
> > +                                  string table pred point)))
> > +              (and probe (cons probe style))))
> > +          (completion--styles metadata))))
> > +    (completion-adjust-metadata-for-style (cdr result-and-style)
> metadata)
> >      (if requote
>
> But here it appears you're not using the return value, unless I'm
> missing something.
>
>
>         Stefan
>
>

[-- Attachment #2: Type: text/html, Size: 2616 bytes --]