I’m running Xfree86 4.3.0 (although this behavior is also in 4.2.1)… using
SDL_WarpMouse sometimes (not always) Xlib reports “Xlib: unexpected async
reply (sequence 0x56)” and then waits for as long as 11 seconds before
returning from the library. I’ve seen it return in as few as 2 seconds …
So my query is… is there something I can do to make its return more
expediant? Has anyone noticed a similar problem?
I’m also re-attaching the fb-open-lock patch which has the comments replaced
from // to /* and has removed the additional modifications to the table of
frequencies.
Unfortunatly I’m stuck using ‘official’ releases here at work… and I
definatly NEED this lock to fail silently - otherwise the application enters
a spin-lock and fails to respond to clients in time, which they then decide
that the server is gone, defunct, etc, when switched to another virtual
terminal…
The other patches were convenient for me - but probably won’t work in the
longrun, and I’m also gathering from the general traffic on the list that
most people are targeting X and not frame buffer so there’s little
opportunity to gather input about whether updating the screen from the saved
surface is good, etc…
Friday I had done a cvs update and had noted that the patches hadn’t been
inserted… (and I do understand that other people in the world are just as
busy as me )
this should be a new thread. People shouldn’t reply to other messages on the mailinglist to start a new thread. Please learn.On the subject of your problem, I suspect that you are doing SDL things in the wrong thread, or in a timer callback. You should only call (most) SDL functions in the main thread… LIM- Jim wrote:
I’m running Xfree86 4.3.0 (although this behavior is also in 4.2.1)… using
SDL_WarpMouse sometimes (not always) Xlib reports “Xlib: unexpected async
reply (sequence 0x56)” and then waits for as long as 11 seconds before
returning from the library. I’ve seen it return in as few as 2 seconds …
So my query is… is there something I can do to make its return more
expediant? Has anyone noticed a similar problem?
I’m also re-attaching the fb-open-lock patch which has the comments replaced
from // to /* and has removed the additional modifications to the table of
frequencies.
Unfortunatly I’m stuck using ‘official’ releases here at work… and I
definatly NEED this lock to fail silently - otherwise the application enters
a spin-lock and fails to respond to clients in time, which they then decide
that the server is gone, defunct, etc, when switched to another virtual
terminal…
The other patches were convenient for me - but probably won’t work in the
longrun, and I’m also gathering from the general traffic on the list that
most people are targeting X and not frame buffer so there’s little
opportunity to gather input about whether updating the screen from the saved
surface is good, etc…
Friday I had done a cvs update and had noted that the patches hadn’t been
inserted… (and I do understand that other people in the world are just as
busy as me )
this should be a new thread. People shouldn’t reply to other messages on
the mailinglist to start a new thread. Please learn.
Ahh I see didn’t think about the hidden ‘references’ fields and junk
I’m running Xfree86 4.3.0 (although this behavior is also in 4.2.1)…
using
SDL_WarpMouse sometimes (not always) Xlib reports "Xlib: unexpected
async
reply (sequence 0x56)" and then waits for as long as 11 seconds before
returning from the library. I’ve seen it return in as few as 2 seconds
…
So my query is… is there something I can do to make its return more
expediant? Has anyone noticed a similar problem?
I’m also re-attaching the fb-open-lock patch which has the comments
replaced
from // to /* and has removed the additional modifications to the table
of
frequencies.
Unfortunatly I’m stuck using ‘official’ releases here at work… and I
definatly NEED this lock to fail silently - otherwise the application
enters
a spin-lock and fails to respond to clients in time, which they then
decide
that the server is gone, defunct, etc, when switched to another virtual
terminal…
The other patches were convenient for me - but probably won’t work in
the
longrun, and I’m also gathering from the general traffic on the list
that
most people are targeting X and not frame buffer so there’s little
opportunity to gather input about whether updating the screen from the
saved
surface is good, etc…
Friday I had done a cvs update and had noted that the patches hadn’t
been
inserted… (and I do understand that other people in the world are just
as> On the subject of your problem, I suspect that you are doing SDL things in the wrong thread, or in a timer callback. You should only call (most) SDL functions in the main thread… Yeah - very likely in fact - will fix that and expect that to help thanx. LIM- Jim wrote:
)