CEDET

Hi all,

  For those who do incremental updates from CVS, you will need to
delete eieio/eieio-comp.elc to get the build working.  Some CLOS
compatibility issues have been found, and a case-change for some of
the keywords has been made.  The old byte-compiler support for eieio
classes will fail to compile eieio.el, so the deletion is needed.

Enjoy
Eric

-- 
          Eric Ludlam:                       eric@...
   Siege: http://www.siege-engine.com          Emacs: http://cedet.sourceforge.net

Hello all,

  Last week I had checked in a feature based on an email discussion
which was not CC'd to the list.

  A common feature question was, for C/C++ why semantic-ia-fast-jump
always goes to a prototype, and not to the implementation.  The reason
is because of the way searches are optimized to only the include file
list.  Trying to find the implementation for a prototype turned out to
be a bit tricky to get right, especially with namespaces and such
involved.

  I've put a new analyzer library together to try and solve this
problem.  I added a very simple user interface to with the command
semantic-analyze-proto-impl-toggle.  Bind this to a key, and press it
to jump between a prototype and its implementation.

  It works ok in my test directories, but I haven't had much
opportunity to test it beyond that.  If it doesn't work the first
time, you may need to try `semanticdb-dump-all-table-summary' and make
sure the right directories and tables had been loaded.  The idle work
tasks should automatically pick up all the files needed, but it
depends on the breadth of your project.

  If it still doesn't work, let me know with any repro steps, and we
can start working out how to debug and get this thing working
correctly.

Thanks
Eric

-- 
          Eric Ludlam:                       eric@...
   Siege: http://www.siege-engine.com          Emacs: http://cedet.sourceforge.net

Hello all,

I'm wondering if there is a way to get speedbar to display entire
function signatures on a single line.  The default behavior is to put
the function name as one node, its return type and argument names as
children, and the type of each argument as grandchildren.  I am using
speedbar in a dual-head configuration, so I am using one monitor for
speedbar and one for the edit window.  As a result, I have lots of
horizontal space for speedbar, whereas it seems to be rather
compressed horizontally.  I'm trying to decompress it a bit, and get
more information with less navigation.

thanks
-Matt

>>> "Eric M. Ludlam" <eric@...> seems to think that:
>>>> Gerdin Tobias <Tobias.Gerdin@...> seems to think that:
  [ ... ]
>>>  There is a way to get the definition of a class, and that is
>>>via the "brutish" search.  Getting that to work requires
>>>parsing every file in your project.  There are likely clever
>>>ways to get it to associate a header file with some group of C
>>>files, but that would not be reliable.
  [ ... ]
>I tried my hand at go-to-definition this morning, and I came up with
>the attached fcn.  Sadly, my version depends on a semanticdb API I
>haven't finished checking in yet, so I put in a hack to make it work
>w/out.  As the todo comment also suggests, there is the potential for
>confusion if you use the same name a lot, so this needs some work.
>
>Enjoy
>Eric
>
>(defun senator-jump-to-definition ()
>  "Jump to the definition of the tag under point.
>This does not jump to the symbol the cursor is on, but the entire
>tag (function, type, etc)."
>  (interactive)
>  (let ((tg (senator-current-tag)))
  [ ... ]

The function I gave above might work well for plain functions, and
sometimes work for methods, but I found that calculating an accurate
match of where to jump to was surprisingly complex.  Like, way
complex.

Now in CVS, there is `semantic-analyze-refs.el' for trying to analyze
all these references accurately when classes and namespaces are
involved.

The simple seeming function `semantic-analyze-proto-impl-toggle' will
bounce between an implementation and a prototype.  I bound it to F8
for a while, and found it fast in my test directories.  I haven't
tried it on a real code base.

Give it a try, and let me know how it goes.  There is yet a lot of
room for speed optimizations here, so exercising it is the first step.

Eric

-- 
          Eric Ludlam:                       eric@...
   Siege: http://www.siege-engine.com          Emacs: http://cedet.sourceforge.net

hello all,

I've noticed that sometimes, when I switch buffers using C-x b,
speedbar automatically changes the directory it's displaying.  Is
there a way to turn this behavior off?

thanks
-Matt Brown

Hi all,

  I've been running through the PRE_RELEASE_CHECKLIST for CEDET, and
have been finding lots of subtle issues to deal with, putting the next
pre-release on hold for a short while.  Plus, as usual, I get
distracted easily by shiny new concepts that need implementing.

  I updated the CEDET load system to not display so many messages.
Displaying messages is apparently time-consuming, so this is a minor
startup speed boost.  This also involved updates to inversion.

  The status progress bar "working" had a bug David Engster sent me a
patch for.

  I discovered a subtle bug in EIEIO multiple-inheritance method
invocation.  It was doing a breadth first search for the first method
in a MI case, and if the second inheritance tree was shorter than the
first, its method would be called first.  Extra sneaky issues there.

  EDE has seen some major bug fixes.  The project-am.el system has
been enhanced, to include some more known Automake variables, and I
fixed bugs in the way that the LISP, TEXI, and MAN types are read from
the Makefile.am.  I now remember how it works, so if you use the raw
Automake support, and know of some things that are missing, let me
know and they can be added easily.  This was needed to get fully
automated include paths working in a sample project I have.

  I also created an Emacs simple project type.  If you hack Emacs'
source, then EDE will now automatically identify it as a project, and
setup Semantic's include paths, and the all important CPP macro table,
such that DEFUN works.  Huzzah!  If you hack Emacs C code, you want
this update.

  This also involved fixes to the EDE speedbar support, and the EDE
doc has had some updates too.

  Getting Emacs sources to parse required changes in the C parser to
handle a new case (for me) for the old style knr arguments.

  There have been some subtle updates to the semantic analyzer, and
analyzer debugger.

  semantic-doc has a minor fix, and the speedbar analyzer smart
completion thingy now displays function doc for the symbol the cursor
is on.  In C/C++, the "doc" is a comment that shows up just before
your symbol.  It's interesting, try it out:
  M-x semantic-speedbar-analysis

  A nice enhancement might be to make it also work w/ comments on the
same line after a definition.  That's now somewhere on my todo list.

  The way that semanticdb reparses files not in a buffer has changed,
and more features are turned off during the temporary parsing.  Each
new disabled feature makes it faster.  Check
`semantic-find-file-noselect' to see if it should be disabling some
feature you use, allowing it to go faster.

  The javascript parser and db have been moved from contrib to
semantic proper.

  There is a new format function `semantic-format-tag-short-doc'.  You
can use this in place for the idle-summary-function, or any other tag
formatting routine.  Likely in ECB tree-view too.  I wonder what
that's like?

  There were also a multitude of small bug fixes, and lots of doc
updates and fixes.

  If you use MATLAB, the matlab-emacs project has some new CEDET
support in it as well.  Check it out.

  As some fairly basic structural things in the build system, and in
EDE have changed, it would be great if you could update from CVS, as
I'd like to hear if there are any issues with these updates.  Bug
reports and patches welcome.

Thanks
Eric

-- 
          Eric Ludlam:                       eric@...
   Siege: http://www.siege-engine.com          Emacs: http://cedet.sourceforge.net

Hi,

I always thought that :initform stored in class-public-d slot
of the vector class was inherited. I wonder how I never
encounter the problem. But I discover now that it is not the
case. Indeed, value is overriden when defining a subclass even
if nothing is specified in the subclass. In this case slot
become unbound by default. Should we not make a simple test
inhibiting overriding when nothing is specified in the subclss
keeping the initform of the superclass ?

Regards

Pierre

Hi,

On 2008-08-26 02:56, Eric M. Ludlam wrote:
> A good thing to try when percieved completion doesn't work is:
> 
> C-u M-x bovinate RET
> 
> which reparses the local buffer, and resets various caches.  Then use
> a completion function bound to a key, like C-c , TAB.
> 
> Does this sound like your case?  I was not able to replicate your
> issue w/ (global-ede-mode 1) in my .emacs file.

That sounds like my case but using bovinate doesn't always helps. If
global-ede-mode is activate in .emacs I can run C-u M-x bovinate RET
before or after the first completion and it won't work.

If however global-ede-mode is _not_ activated in .emacs starting
bovinate right after M-x global-ede-mode works and it will complete the
symbol FOO.

I have tested it with a stripped-down .emacs file but that doesn't make
any difference.

I also changed

  (semantic-load-enable-code-helpers)

in my .emacs into

  (semantic-load-enable-guady-code-helpers)

but it doesn't change anything except that it paints the "test.h"
include line red when doing the first completion (if global-ede-mode is
deactivated in .emacs). After activated global-ede-mode and doing
"bovinate" it's no longer red and completion works.

With global-ede-mode activated in .emacs this line isn't red but
completion doesn't work. I will do some more test to find out what
triggers this problem on my system.

Best Regards,

Ralf Hoffmann

-- 
Ralf Hoffmann <ralf@...>
Homepage: http://www.boomerangsworld.de

Hi,

  I haven't seen this, likely because I enable srecode-minor-mode, so
everything is loaded before EDE runs.

  As such I've added a require before the use in ede to make sure it
is aviailable.  Hopefully this will solve your problems.

Thanks
Eric

>>> Pierre Lorenzon <devel@...> seems to think that:
>
>Hi,
>
>I upgraded cedet to the last cvs version and then try to
>execute ede-project-regenerate. And I got following error :
>Debugger entered--Lisp error: (void-variable
>srecode-map-load-path)
>
>Indeed, this variable srecode-map-load-path seems to be defined
>in srecode-map.el and maybe this library is not
>required. Is it a bug ? Should we patch that ? I can try to
>explicitely require srecode-map but it is maybe not the most
>elegant manner ! 
>
>Regards
>
>Pierre
  [ ... ]

-- 
          Eric Ludlam:                       eric@...
   Siege: http://www.siege-engine.com          Emacs: http://cedet.sourceforge.net