[cairo] AC_TRY_RUN vs crosscompile

[Uli Schlachter]

On 15.04.2017 18:08, Enrico Weigelt, metux IT consult wrote:
> On 15.04.2017 12:55, Uli Schlachter wrote:
> 
>> The race condition now is: If cairo deletes the segment before the X11
>> server attached, the server will fail to attach. If the server is
>> faster, things will work out.
> 
> So, there's no way finding out when the server has attached it ?
> After the XSync call ? What is the cost of that extra call - is that
> really notable ?

One round-trip to the server (involving a context switch, likely).

I don't know the performance impact of this precisely, but given that
someone went to the trouble of writing all of this instead of just
always doing an XSync(), I guess it there was a reason for this.

> Can't we just move the remove call to a place where we know that
> the server has attached it ?

Cairo's Xlib backend tries hard to avoid blocking requests, so it's hard
to figure out when the server really is done. Plus, that requires
additional memory to save this information that is not needed for
something else.

So, no idea where that place could be exactly.

Uli
-- 
"Every once in a while, declare peace. It confuses the hell out of your
enemies"
 - 79th Rule of Acquisition