[cairo] Engaging 64 bit precision

[Richard Procter]

On 12/08/2010, at 9:14 AM, Richard Procter wrote:

> Hi there,
>
> I'd like to use a 64 bit type in cairo's path calculations and at  
> least 30 bits of fractional precision. This looks tantalizingly  
> possible going by the defines in cairo-fixed-type-private.h.
>
> Has anyone already done this or have any pointers on how to go  
> about it?
>
> I see there's a warning that CAIRO_FIXED_BITS "must be 32" as the  
> tesselator makes assumptions about the fixed type, however I don't  
> believe I'm using it as I am using/abusing cairo only for its clean  
> path construction API and python bindings (having pdf debug output  
> in development has also been very handy, too).


FWIW, here's is my beginner's successfully compiled but untested  
first attempt to boost CAIRO_FIXED_BITS to the maximum 64 bit double  
precision of 52 bits.

It causes GCC to emit a few integer overflow warnings in cario- 
analysis-surface.c:{807, 808} and cairo-pattern.c:{2283,2284}.

Comments welcome.

regards,
Richard.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: 52bit.patch
Type: application/octet-stream
Size: 5774 bytes
Desc: not available
URL: <http://lists.cairographics.org/archives/cairo/attachments/20100812/980b4cf6/attachment.obj>
-------------- next part --------------






>
> Has anyone already done this or have any pointers on how to go  
> about it?
>
> I see there's a warning that CAIRO_FIXED_BITS "must be 32" as the  
> tesselator makes assumptions about the fixed type, however I don't  
> believe I'm using it as I am using/abusing cairo only for its clean  
> path construction API and python bindings (having pdf debug output  
> in development has also been very handy, too).
>
> many thanks,
> Richard.
>
>