Re: Annotating completion of labels

[Colin Baxter]

Dear Arash,

Thanks for your response
>>>>> Arash Esbati <arash@gnu.org> writes:

    > Colin Baxter <m43cap@yandex.com> writes:
    >> Ok, I suppose for people who like that sort of this.

    > Of course, the whole in-buffer completion thingy is for people who
    > like it.

    >> If implemented then I hope it would be off by default.

    > The in-buffer completion machinery in AUCTeX is available by
    > default, it kicks in once you use it.  It doesn't get in the way
    > if you don't use it.  And in general, the related functions are
    > designed to return quickly.  And file size doesn't really matter,
    > other that during parsing.

    >> I have many hundreds of labels in some TeX files and I would
    >> worry that what you're proposing might slow down or hang my
    >> emacs.

    > If you're using RefTeX you shouldn't notice a difference.  My
    > implementation uses `reftex-access-scan-info' which RefTeX does as
    > well once used.

    >> In any case, I generate random numbers to insert as labels, so
    >> completion would be meaningless I think.

    > Maybe I'm missing the point, but this is about referencing labels
    > with \ref et al.  So why is completion meaningless inside a
    > referencing macro?  I admit that RefTeX itself offers a much more
    > powerful interface with `reftex-reference' bound to 'C-c )'

I think it might be me who's missing the point.

When I insert a \label then what's in the bracket is always a random
alphanumeric, as in say \label{58f7161e}, and never some descriptive
text. Obviously I refer to the label with \ref{58f7161e}. This is done
by hand; something which I don't find inconvenient because I think it is
more important for me to know where and what is written in the text
rather than ease of use.

I envisaged that using your completion function would give me a choice
of random numbers. Even if some run-on texts is included in the
completion buffer then this still might not provide sufficient context
for me to choose the completion accurately, especially if \label is some
distance away in the text from the corresponding \ref. All this might
seem inconsistent with my supposed "knowing the text" and I may have
completely misunderstood what you propose.

Best wishes,

Colin Baxter.