Re: [PATCH] Handle case where `beg` and `end` are strings instead of markers

[Stefan Monnier <monnier@iro.umontreal.ca>]

James N. V. Cash [2022-04-30 08:41:24] wrote:

> Stefan Monnier <monnier@iro.umontreal.ca> writes:
>
>>>> +  (let ((beg (if (number-or-marker-p beg)
>>>> +                 beg
>>>> +               (save-excursion
>>>> +                 (goto-char (minibuffer-prompt-end))
>>>> +                 (search-forward beg))))
>>>> +        (end (if (number-or-marker-p end)
>>>> +                 end
>>>> +               (save-excursion
>>>> +                 (goto-char (point-max))
>>>> +                 (search-backward end)))))
>>>
>>> What if there are several matches for `beg` and/or for `end` in
>>> the buffer?  How do we know we got the intended one?
>>
>> IIUC part of the purpose of using strings rather than positions is to
>> account for the case where the text was modified between the moment the
>> *Completions* list was created and the moment the user selects one of
>> the options.  So another question is what to do if the search fails (the
>> above code signals an error in that case, which is probably not what we
>> want).
>
> That is a good question. That would happen, I guess, if you had selected
> multiple candidates in a CRM, started completing, then deleted the
> previously entered candidates, then selected one. I suppose I'd expect
> it to add the completed candidate to the end of the remaining list?
> So something like
>
>   (let ((beg (if (number-or-marker-p beg)
>                  beg
>                (save-excursion
>                  (goto-char (minibuffer-prompt-end))
>                  (or (search-forward beg nil t)
>                      (search-forward-regexp crm-separator nil t)
>                      (minibuffer-prompt-end)))))

This makes it specific to CRM.  I think we should try and think more
locally, without paying too much attention about the various possible
external cases: what do we want the `beg` and `end` strings to *mean*?

Maybe instead of strings, they should be functions, so CRM can set them
to functions that first look for the separator and then search for some
leading/trailing string?


        Stefan