Re[1]: [CEDET-devel] parsers' plug-in API

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

>>> pon...@ne... (David Ponce) seems to think that:
  [ ... ]
>>  Lastly, I thought it would be useful to build an EIEIO class to
>>handle modes and overrides.  This would bring a more traditional OO
>>framework to how parsers are specialized.  This would make some
>>aspects easier to program, and is probably a longer term goal.
>
>I agree with you on this point even if it surprises me a little bit!
>If I correctly remember, sometimes ago you told me that you didn't
>want to use EIEIO in semantic's core, because that would probably
>prevent a future inclusion of semantic in GNU/Emacs!  Have you changed
>your mind?
>
>As I talk of inclusion in GNU/Emacs you might be interested to know
>that I signed a global copyright assignment for my work on GNU/Emacs
>:-)  So there shouldn't be any problem if you still have plan to
>include [parts of] semantic in the official distribution.

There were two things I was worried about with EIEIO.  One was, as you
say, adding a dependency that might prevent it being distributed in
GNU Emacs.  Also, RMS doesn't like core tools in Emacs depending on CL
macros.  As RMS is also interested in having EIEIO be a part of Emacs,
that eliminates one worry.  As I already used EIEIO in semanticdb, I
think the programatic benefits outweigh (for myself) any style opinion
RMS may have.  I may also just be that I'm getting ornery.

The second reason was that I didn't want EIEIO macros in what needs to
be high-performance code (ie, as a way of defining tokens, or
executing methods in the parser.)  I still think this is true.  I
think EIEIO would be best used as a way of wrapping up high level APIs
where the number of iterations over the EIEIO wrappers is minimal.

>[...]
>>  If it is relevant, could you put this into the 1.4 version of
>>semantic.el?  Thanks.
>
>Of course!  I will do that as soon as I had tested the new code a
>little bit more.

I was referring to just the one bug-fix you mentioned in your previous
patch.  I did eventually look at it yesterday, but it was unclear to
me if it translates into the old code.

I think the new API should wait till after 1.4.  I checked in some of
the doc files (like INSTALL and semantic.texi) redirecting help
requests to go to the mailing list.  Hopefully that will ease up on
the amount of email I get.  I still have a pile of patches to the c
parser to add in.  I'll make an extra effort to make that tar ball.

Eric

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