>> I suppose that you have some stuff like "Auto rebuild other buffers" (menu
>> "Senator --> imenu -->..." active .... Check the semantic-manuals...

>You should set semantic-imenu-auto-rebuild-directory-indexes to nil
>for this feature.   Does anyone have an opinion if this should be the

Yes, concerning this topic i have a very clear-cut opinion: The default
value should be nil, if not nil then nil, if not nil then nil ;-)))
Each other default-value triggers a lot of complaints abouts slowdowned

>> >
>> > IMO, such indexing tasks should really run in their own thread. I
>> > don"t
>> > know if there is an easier solution to my problem or if the current
>> > setup would allow a threaded approach easily, but that"s the way to go
>> > about it in the long term really.
>> Problem: Emacs does not support threads for several tasks!! It just support
>> running some time-consuming tasks in idle-times...it is important, that this
>> tasks are interruptable via arbitrary keyboard-hit or mouse-click... AFAIK
>> Most of semantic-jobs are...Eric?
  [ ... ]

>I've made as much of the idle-time stuff interruptible as is
>feasible.  I had not made the imenu directory thing interruptible so
>that could be suspect.

>As Klaus said, Emacs does not support threads.  Instead we have to
>multiplex around user actions.  I think the CVS version is much more
>responsive that older 1.x versions.  The new incremental parser also
>is very fast.

Yes, i can confirm that - much better! But i think we could make the
interruption even more better - at least with GNU Emacs... the CVS Emacs
(22.X) contains a new macro `while-no-input' which allows now to write
each task interruptable even if such a task runs a longrunning os-command.
If i have understood this new `while-no-input' right it is much better
then writting something like

(while (not (input-pending...