On Tue, Mar 21, 2000 at 04:37:11PM +0000, Nick Ing-Simmons wrote:
> Graham Barr <[EMAIL PROTECTED]> writes:
> >> If I understand the code, pe_sys_multiplex basically waits in a poll/select
> >> call until something pops up to wake it up. This implies that there is
> >> essentially only one waiting state, which all watchers have to tie into. In
> >> other words, if Event is sleeping on a set of filehandles, and some other
> >> type of event comes in (an X event handled by a new type of watcher?) the
> >> poll/select won't wake up and the event won't be serviced immediately.
> >
> >But the X event would enter the application on a file descriptor.
>
> True, one does do a select/poll on the X socket(s).
> But you need to layer another "logical" watcher on top of that
> as one "message" up the socket may contain multiple X events
> (Xlib sucks all the XEvents out of the socket and queues them)
> so "socket not readable" does NOT mean "No XEvents pending".
> Also IIRC this is TCP socket so an XEvent may not be complete.
> You also need to XFlush() pending outgoing requests prior to the select().
Yes, but when the X socket becomes readable you ask Xlib for the events
and queue them in Event's queue. This will cause them to be processed
before sleeping again.
Graham.
