CEDET

> A while back, we had a discussion about semantic token formats, and
> concocting a single function for constructing them.  Then both parsers
> could use them to create tokens, and once that was safe, we could
> change the underlying storage. (If desired).
> 
> I'm going to take a first stab at this.  Eventually, I'll check in
> semantic-token.el with elements from semantic-fw and semantic-util in
> it.  After that, I'll build the constructors.  Then we can look at
> porting old parsers to the new constructors.
> 
> Thoughts=3F
> Eric
> 

Will you use EIEIO objects to represent tokens or a simpler (more
efficient=3F) property list approach, or a different mechanism=3F

Anyway, that would be a great enhancement, that would probably require 
important changes to the existing code ;-)

David

A while back, we had a discussion about semantic token formats, and
concocting a single function for constructing them.  Then both parsers
could use them to create tokens, and once that was safe, we could
change the underlying storage. (If desired).

I'm going to take a first stab at this.  Eventually, I'll check in
semantic-token.el with elements from semantic-fw and semantic-util in
it.  After that, I'll build the constructors.  Then we can look at
porting old parsers to the new constructors.

Thoughts?
Eric

-- 
          Eric Ludlam:                 zappo@..., eric@...
   Home: http://www.ludlam.net            Siege: http://www.siege-engine.com
Emacs: http://cedet.sourceforge.net               GNU: http://www.gnu.org

Hi Richard,

> I thought it might be good to move all the documentation from the old
> semantic.texi to approprate chapters in the new manual first before
> starting to add new stuff.  I could do that unless you or David would
> like to get to it first.  Let me know.

That would be great.  Please do.

Thanks!
David

Hi Eric,

> There is a routine in semantic-util that finds externally defined
> children for a given type.  This routine really bogged things down
> with the Emacs Lisp system db.
> 
> The core routine for that is now in semanticdb-search and can be
> overriden by specific types of system databases.  I removed the old
> code from semantic-util.  I also changed semantic-analyze to use the
> new routine.  While I was in semantic-util, I found some more
> functions that needed to be converted to `define-overload'.
> 
> I also found an fixed some bugs in semantic-analyze.  The analyzer
> mode for speedbar is now much more robust for C++, and significantly
> more accurate.  If you use C++, give it a try.

All that looks good.  Thanks!

Maybe it would be worth adding your comments below I found in
semanticdb-search.el, in the NEWS file=3F

;; There are three types of searches that can be implemented:
;;
;; Basic Search:
;;  These searches allow searching on specific attributes of tokens,
;;  such as name or type.
;;
;; Advanced Search:
;;  These are searches that were needed to accomplish some tasks
;;  during in utilities.  Advanced searches include matching methods
;;  defined outside some parent class.
;;
;;  The reason for advanced searches are so that external
;;  repositories such as the Emacs obarray, or java .class files can
;;  quickly answer these needed questions without dumping the entire
;;  symbol list into Emacs for a regular semanticdb search.
;;
;; Generic Search:
;;  The generic search, `semanticdb-find-nonterminal-by-function'
;;  accepts a Emacs Lisp predicate that tests tokens in Semantic
;;  format.  Most external searches cannot perform this search.

David

Eric,

I thought it might be good to move all the documentation from the old
semantic.texi to approprate chapters in the new manual first before
starting to add new stuff.  I could do that unless you or David would
like to get to it first.  Let me know.