Subject: Re: Commit: back out martin's changes ;-(
From: Dom Lachowicz (dominicl@seas.upenn.edu)
Date: Wed Sep 05 2001 - 12:44:01 CDT
Quoting David Mandelin <mandelin@cs.wisc.edu>:
> Yes, but the timers could be changed so that they don't call functions
> directly, but post "timer" events instead. Similarly, the event loops
> could be set up to generate "idle" events every time AbiWord goes idle.
> The event mapper would take care of the rest.
This is a good suggestion. What I don't want to see: an idle put into the
UT_Timer class. I'd rather see a separate UT_Idle class for this *if we choose
to go through this route*.
Having events get posted to our event queue is a good suggestion. However, we
still need some protection for the event queue itself, since the timers still
interrupt. eg:
Inserting event into position 2 in queue
Timer fires, sees position 2 open in queue
Timer inserts event into position 2 to queue a spell-check
Timer exits
Original process enters its event into position 2 in the queue
Result: 1 event gets lost, and the word never gets spell-checked.
Still bad news, I'm afraid, but definitely fixable.
Dom
This archive was generated by hypermail 2b25 : Wed Sep 05 2001 - 12:44:04 CDT