bug#20365: 24.5; all-completions returns duplicates for Info-read-node-name-1

[Oleh Krehel <ohwoeowho@gmail.com>]

On Mon, Apr 20, 2015 at 4:38 PM, Stefan Monnier
<monnier@iro.umontreal.ca> wrote:
>> nodes. This concept can be adapted to all compltetion types with
>> "infinite" candidates. And there would be zero confusion:
>> `all-completions' would immediately return a list of strings, some of
>> them terminal nodes, some of them "directories".  Then it only remains
>> to provide a generic `file-directory-p' and we're done.
>
> Yes, that's also what I was thinking.  Basically, if all-completions
> returns something which requires further input, then this something
> should include the "terminating char" which makes completion-boundaries
> change (which is how to detect that your candidate list is out-of-date
> because the input has "moved to another directory").

I'm not fully familiar with the concept of completion-boundaries, but I
have a feeling that I won't like it, specifically the multiple callback
passes for a response to a single completion input change.

Is the concept of completion-boundaries irreplaceable, could it maybe be
replaced by a notion of navigating a tree (just like navigating a file
system, with terminal and non-terminal nodes).  Which Emacs features use
the completion-boundaries concept?

>>> That's the case currently.  The completion-table is called and the
>>> caller is the UI, and currently it's the UI's responsibility to remove
>>> the duplicates.
>> So Info returning duplicates is a bug that should be fixed?
>
> Our UI already does remove duplicates (not in info.el, of course, since
> our UI is in minibuffer.el).

I've added a fix to Info in any case. Can you please look at it in the
fix-info-dups branch?

>>>> Here's my line of thought: a completion function is expected to have
>>>> an O(N) complexity, where N is the amount of candidates.  Removing
>>>> duplicates is O(N^2) at worst, and O(NlogN) at best.
>>> Actually, with a hash-table it's pretty much down to O(N).
>> Yeah, but we're not using that.
>
> Not sure who's "we", here.  But the point is that if the performance of
> delete-dups becomes a problem, it can be improved.

We are the happy Emacs users.

>> And having no assumptions on the data, the hashing function would be
>> the most basic one.
>
> I don't think that should make much difference: (make-hash-table :test #'equal)
> should work just fine.

OK, that's good to keep in mind. But even better is to avoid placing the
overhead on the UI, be it minibuffer.el or ivy.el or whatever.

Oleh