Bug#553500: blank window workaround patch works only by accident

[Vincent Legout]

Hi Monty,

On Wed, Mar 23, 2011 at 06:05:09AM -0400, Monty Montgomery wrote :
> I have a new 'release' of gPlanarity with a few new translations, and
> a correct fix to the blank window bug.  See:
> http://web.mit.edu/~xiphmont/Public/gPlanarity.html
> 
> The fix for this bug suggested in this bug report (553500) is not
> correct and works by accident, meaning it could also break at any
> time.
> 
> The bug is caused by a restriction/clarification in the GDK/Cairo
> interaction API where a cairo target created for a given GDK window is
> now only gauranteed to be valid for the span of a single GDK event.
> gPlanarity had previously been caching a single cairo_t upon
> allocation of the window and attempting to use it for all following
> drawing operations.  It now allocates a cairo_t for the GDK window on
> demand for each drawing op, then discards it.  That is the correct fix
> to the bug.

Yes, I know my patch was not ideal. Thanks for fixing this bug properly, I'll
prepare a new package for Debian soon.

Vincent