>>> Ralf Hoffmann <ralf@...> seems to think that:
[ ... ]
>Now I have another problem: The completion doesn't (always) know about
>symbols from my include directory. In the example I sent there's a
>symbol FOO in include/test.h but it will not be completed in src/main.c
>unless I copy it to src.
>The interesting thing is that it works under some circumstances:
>If I have (global-ede-mode t) in my .emacs it doesn't work.
>If (global-ede-mode t) is commented out it works if I open the file and
>start global-ede-mode manual _before_ doing any completion. If I start
>global-ede-mode after a completion it also doesn't work.
[ ... ]
I can almost picture how this might happen.
If you start w/ ede on, and try to complete in main.c the first time
by using the idle completion tool, the test.h may not have been parsed
If you restart w/ ede off, then turn it on, there may have been some
pause or wait that allowed semantic to switch into "resolve all the
headers mode", thus enabling completion in idle time.
If you leave EDE off, then edit the files, then turn it on, it will
have already resolved the locations of the headers, and then it won't
try to resolve again once EDE is turned on.
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.
Eric Ludlam: eric@...
Siege: http://www.siege-engine.com Emacs: http://cedet.sourceforge.net