Re: antialiasing for emacs

[Chris Gray <christopher.gray@mail.mcgill.ca>]

On Mon, 28 Jul 2003, Stefan Monnier wrote:
>> Chris Gray <christopher.gray@mail.mcgill.ca> writes:
>> > Yep.  It happens when I start gnus too.  I really don't know why,
>> > and these "async errors" seem very hard to debug.
>> 
>> Hmmm, actually, come to think of it, there's a function to make
>> emacs use X `synchronously', which might make debugging easier...
>> 
>>    x-synchronize is a built-in function.
>>    (x-synchronize ON &optional DISPLAY)
>> 
>>    If ON is non-nil, report errors as soon as the erring request is
>>    made.  If ON is nil, allow buffering of requests.  This is a
>>    noop on Mac OS systems.  The optional second argument DISPLAY
>>    specifies which display to act on.  DISPLAY should be either a
>>    frame or a display name (a string).  If DISPLAY is omitted or
>>    nil, that stands for the selected frame's display.
>> 
>> -Miles
> 
> I don't think it's going to help.  The `async error' is generally
> due to missing BLOCK_INPUT statements that lead to Xlib functions
> being called from the signal handler while we're already inside
> an Xlib function.

This is exactly what the problem was.  In my patch you will notice
that I had BLOCK_INPUT and UNBLOCK_INPUT commented for some strange
reason.  Uncommenting it made the async errors go away.

I want to fix some other things before sending another patch, but
those who are already testing my first patch should know what to do. 

Cheers,
Chris