[CEDET-devel] Recent check ins

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

Hi all,

  I recently checked in two key changes into semanticdb.
Change 1 made the "tables" slot on the database object protected.
I had to do this because you cannot override "oref", and the tables
slot has grown past it's old non-static nature.

  A side effect of this is that many other files which referred to
this slot now must use the :accessor to get the list of tables.  This
is a minor change in those places.

  Some implementations of system override DBs override the :accessor
so that if there are no current tables, one can be created if needed,
or nil can be returned.  This avoids unbound slot errors.

  The second change was to add the semanticdb search results API.  It
is the same API I posted for Klaus a while back.  I have not
integrated his updated ideas in it yet.  Why coming next.

  All these changes are needed to continue upgrading the semantic
completion API.  The previous API only handled single-buffer
completion.  I've extended it to handle completing symbols from an
entire project.  This is why the :table :accessor is needed.  Its
new internal format is a semanticdb search result format, so that API
was needed too.

  Unfortunately the completion code I checked in doesn't work quite
right yet.  I should have accommodated semanticdb from the
beginning.  Adding it on second is a real pain.  Anyway, that stuff
has been waiting for beta1 to be over before I checked it in, so I
have.

  I'll continue working on the completion engine till it is more
stable.  Once that is done I can start replacing existing prompts
(from semantic-util.el, and in cogre) with the new prompts.

  After I finish this (or when I don't want to look at it anymore), my
second task is building a semantic idle timer to replace auto-parse
mode.  Programs like eldoc need a freshly parsed buffer, but the
timers don't always line up.  I want to get auto-parse mode, partial
reparsing, and programs like eldoc to be better integrated.  This
will fix a bug reported earlier where eldoc was forcing reparses on a
buffer with broken list syntax and preventing useful editing.  We may
need to stop using eldoc as a harness. :(

Anyway, have fun
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