Code completion

[Carsten Pfeiffer]

On Thu, Jan 13, 2000 at 10:54:11PM +0100, Benoit Cerrina wrote:

Hi,

> this is probably a nice base, but I'm not sure this can be used to gain the
> same level of completion of what exist for the MS tools (I never tried the
> Symantec or Inprise so I can't tell).

actually, I don't know any of them :-}

> In those tools if you write a '.' after a variable name for example the tool
> (VC++ VJ++ or VB) knows the type of the variable and can propose a list of
> syntactically correct completion, not only one that existed previously.

That would be possible with Rich's namespaces-handling code. You would
probably have multiple KCompletion-objects, one for each namespace.
 
> Maybe it could work if Kwrite detected the variable declaration and at this
> time automatically inserted in the KCompletion dictionnary all the allowed
> completions.

Yes, something like that. I don't know how much Rich's code does, but
that's about how it should work. There has to be a parser, that knows
about all possible namespaces. For every namespace, you need one
KCompletion-object and you should know which variable belongs to
which namespace (well, it's a little more difficult with inheritance).

But at least for me, basic completion-features would suffice, [x]emacs
doesn't do any of this, either.

> PS:
> your web page cannot be viewed by kfm

Uh oh... last I had time to edit those pages, kfm didn't even exist :-}
Will update somewhen in the next holidays.

Best wishes,
Carsten Pfeiffer
-- 
http://www.geocities.com/SiliconValley/1632/