Re[2]: [CEDET-devel] completion in semantic

[Eric M. L. <er...@si...>]

>>> David Ponce <dav...@wa...> seems to think that:
>Hi Eric,
>
> >   I've checked in an experimental file `semantic-complete.el' as a
> > proposal to a framework for performing completion.  The test function
> > is simply `semantic-completion-test'.  I've implemented only one
> > completion mechanism which is a minibuffer prompt for selecting a tag
> > out of the current buffer.  The architecture is discussed in the
> > commentary.
>
>Good work Eric!  I think your new completion framework is a good
>idea, and I like the formal distinction between "completor" and
>"displayor".
>
>I did a quick try of `semantic-complete-test' on some buffers and
>noticed the following problems:
>
>- the completion engine don't take into account the current buffer, it
>   only completes based on the first buffer where
>   `semantic-complete-test' was used for the first time.  Even worse,
>   after killing that buffer, completion test no more work and I get
>   this error:

Aha, a typo in the `make-variable-buffer-local' prevented that from working.
  [ ... ]

>- when refining the selection, the completion buffer continues to show
>   all the possible names.  For example when I enter "xxx" in the
>   minibuffer, then press the TAB key twice, I would like to see only
>   the names that begin with "xxx".

Hmm, I see a premature optimization written incorrectly.  I removed
it and if we need a speed boost, I can add something back later.

>- when trying to enter something at first "Symbol:" prompt, I get this
>   error:

Hmm, unbound slots.  I see one possibility I can tidy up.

I'll check these changes in.

Eric

-- 
          Eric Ludlam:                 za...@gn..., er...@si...
   Home: http://www.ludlam.net            Siege: www.siege-engine.com
Emacs: http://cedet.sourceforge.net               GNU: www.gnu.org