Okay, here is what I have found so far:
 
The problem only occurs when I set semantic-load-turn-useful-things-on to true! I grepped cedet for this variable and found it used in semantic-load.el at the very end of the file. Here I find a (when ...) statement which has a variety of statements inside of it. I have been able to narrow it down to the relatively bare minimum number of statements required to reproduce this problem (PLEASE SEE below).
 
I have identified (global-semantic-idle-completions-mode 1) to be the "root" of this problem. When I comment this line out, the problem goes away. In order to avoid symbol definition errors at load time, I have to also include
(global-senator-minor-mode 1) and (global-semantic-idle-scheduler-mode 1). Just those two by themselves do not cause the problem. It's only after I uncomment (global-semantic-idle-completions-mode 1), that the issue appears. I further tested by commenting out the offending line only and letting the (when) do everything else but that. The problem no longer happens.
 
There was another mail earlier today that also related to semantic-load-turn-useful-things-on turning on some "unuseful things" as well. :-)  I believe that with that variable on, the user saw the methods buffer go blank some times. He would then have to rebuild in order to see the methods. I have also seen the same problem, but there was never a real resolution to my posts about it.
 
Anyway, IMHO there is some sort of bug/conflict going on here. I look forward to any comments and/or fixes.
Thanks a lot!
 
 

 
(when (or semantic-load-turn-everything-on
   semantic-load-turn-useful-things-on)
 
;;   (if (and semantic-load-turn-everything-on
;;     (fboundp #'which-func-mode))
;;       (add-hook 'semantic-init-hooks (lambda ()
;;            (which-func-mode 1))))
;;
;;   (when (and (eq window-system 'x)
;;       (locate-library "imenu"))
;;     (add-hook 'semantic-init-hooks (lambda ()
;;          (condition-case nil
;;       (imenu-add-to-menubar
;;        semantic-load-imenu-string)
;;            (error nil)))))
;;
;;   (when semantic-load-turn-everything-on
;;     (global-semantic-highlight-edits-mode 1)
;;     )
  (global-senator-minor-mode 1)
;;   (global-semantic-show-unmatched-syntax-mode 1)
 
  (global-semantic-idle-scheduler-mode 1)
 
  (global-semantic-idle-completions-mode 1)
 
;;   (global-semantic-idle-summary-mode 1)
;;
;;   (global-semantic-show-parser-state-mode 1)
;;
;;   (global-semanticdb-minor-mode 1)
;;
;;   (when (boundp 'header-line-format)
;;     (global-semantic-stickyfunc-mode 1))
;;
;;   (when (and (fboundp 'display-graphic-p)
;;       (display-graphic-p)
;;       ;; The above is also asking for Emacs 21... I think.
;;       )
;;     (global-semantic-show-tag-boundaries-mode 1))
;;
;;   (global-semantic-highlight-by-attribute-mode 1)
;;
;;   ;; This loads any created system databases which get linked into
;;   ;; any searches performed.
;;   (if (and (boundp 'semanticdb-default-system-save-directory)
;;     (stringp semanticdb-default-system-save-directory)
;;     (file-exists-p semanticdb-default-system-save-directory))
;;       (semanticdb-load-system-caches))
;;
 )

 -------------
 Javier
 

-----Original Message-----
From: Oviedo, Javier [mailto:joviedo@telogy.com]
Sent: Friday, June 25, 2004 2:00 PM
To: 'cedet-devel@lists.sourceforge.net '
Cc: 'Berndl, Klaus'; 'David Ishee'
Subject: [CEDET-devel] CEDET / echo-area display conflict!!


Hello all:
I originally posted this mail to the ecb list thinking that ecb was the cause of the problem. Since then I have narrowed it down to a semantic. I will follow up with another mail reporting my results, but first I thought I'd just post this as is.



#############################
I'm using ecb-2.2.5 and cedet-1.0beta2b.
I'm using mic-paren.el
#############################

I've just noticed the following behavior, but I'm not sure if it is ecb or cedet related...so let me start with this list first.


I have ecb deactivated and my cursor is directly after the closing paren in the require statement shown below.

(require 'ecb-autoloads)

What I see is the following in the echo-are:

Matches:   (require 'ecb-autoloads) [496]

The (require 'ecb-autoloads) is displayed in the echo-area with the text-properties(font-lock faces) from the original buffer. Everything seems to work fine.



Now I activate ecb. This is where a see something bizarre. When I put my cursor right after that same closing paren, I see the same behavior...at least for a few seconds. After a few seconds, the text-properties seem to disappear (sometimes completely and sometimes partially)...it is no longer displayed with "faces/colors." This is not good. :-)


This only happens when ecb is active. With ecb off I have never seen this nor can I reproduce it.

I have ruled out my ecb settings by removing any customizations.


Thanks.
Javier



 -------------
 Javier