The python support that exists in CEDET right now only needs someone
to go in and "use" it, and supply any tweaks to tune it in. By teaks,
this involves finding the right methods to overload just for python.
(ie, parsing local variables, looking up locally scoped items like using
statements, statement parsing, etc) There is a complete list of things
to do for this on the web page:
for Step 2, 4, and 5. Step 3 (ede support) is needed if you want to
grow beyond a few local files.
A good step is to build some simple examples in the format of the
completion unit tests (semantic-ia-unit.el) where you can run the
completion engine in a controlled way which will make it much easier for
me to assist via email.
On Tue, 2009-08-18 at 05:36 -0300, Fabian Ezequiel Gallina wrote:
> First of all I like to say that I'm really impressed by CEDET and the
> interesting opportunities for developing nice code helpers for Emacs
> this brings (ECB is such a perfect example).
> I'm a python programmer and Emacs Lisp newcomer, I have written myself
> a python autocompletion package for Emacs called gpycomplete using
> pymacs and Emacs Lisp, which basically works as follows:
> 1) It sends the code to the python interpreter
> 2) The python interpreter takes care of executing the code and then
> available completions are calculated by doing introspection on the
> dictionaries where the code is evaluated.
> 3) The python part of the completion package finally returns a list of
> the available completions.
> The problems this thing have is that it is actually executing the
> code, this is perfect in the case you need *really heavy*
> introspection on the objects (it works very much like ipython except
> that the code you write could be at any point) but could be a headache
> when you are doing something which takes a long time or processing.
> So my plan was actually to start analyzing the buffer for clases and
> members of these classes, variables, functions definitions, etc and
> try to avoid the interpreter as much as possible (just let the user
> decide when to use the interpreter by activating a minor mode) and
> this is when I discovered that semantic is actually doing this.
> I understand that to enable "intellisense" support some work is
> needed, I have read on this mailing list several requests for python
> completion support via semantic, so what I would like is to invite
> people who feel who can help to actually create a group in order to
> accomplish this task.
> I really don't have a lot of spare time these days and I don't feel
> that my Emacs Lisp knowledge is enough to accomplish this by myself,
> but anyways, I will eventually start hacking on this, so I would like
> to hear if someone has already started to work on this or if someone
> is interested in helping this come true.
>  http://github.com/fgallina/gpycomplete/tree/master