manually-invoked watchers

2004-04-20 Thread Zefram
I have a need for a type of watcher for which events are only generated manually. I want this for event types along the lines of "redraw required": it's generated as a result of internal program logic, but I want it to be subject to priority queuing. I'm surprised that there isn't already a "non-

Re: playing nicely with signal handlers

2004-04-20 Thread Zefram
Joshua N Pritikin wrote: >Have you tried simply calling ->stop() on the SIGTSTP signal watcher? >I believe that will change the handler back to SIG_DFL. Ah, neat. That gives me option 2. Thanks. -zefram

Oops! (was Re: zombie timers)

2004-04-20 Thread Andy Mortimer
(I thought my mail client had eaten this version, sorry ... *blush*) -- Andy Mortimer [EMAIL PROTECTED] -- Andy walking, Andy tired, Andy take a little snooze -- "Andy Warhol," David Bowie Andy Mortimer writes: > Zefram writes: >> Andy Mortimer wrote: >>>I'm

Re: playing nicely with signal handlers

2004-04-20 Thread Joshua N Pritikin
On Mon, Apr 19, 2004 at 09:18:08PM +0100, Zefram wrote: > Before I started using Event, I had a couple of signal handlers that, > as part of their work, locally modified the handling of their signal. > The interesting case is with SIGTSTP: after doing some application > cleanup, it would temporaril

Re: zombie timers

2004-04-20 Thread Joshua N Pritikin
On Tue, Apr 20, 2004 at 12:00:15AM +0100, Andy Mortimer wrote: > Zefram writes: > > After they've fired, they still exist and can be modified and retriggered. > > See the "again" method. You need to explicitly cancel your watchers. > > Yep, that works, thanks muchly! > > I'm still slightly puzzl

Re: zombie timers

2004-04-20 Thread Zefram
Andy Mortimer wrote: > Since I haven't kept a reference to the >watchers, I can't see that I could call the ->again method on them to >re-trigger them anyway But for the all_watchers method, Event could have held a weak reference to the watcher, allowing it to be GCed