[Bf-committers] OpenCollada issues (was Re: Meeting minutes, feb 14, 2010)

[Arystan Dyussenov]

Your patch fixes the issue. Tested on 64-bit Mac. Cool!

Arystan

On Sat, Feb 20, 2010 at 8:44 PM, Damien Plisson <damien.plisson at yahoo.fr>wrote:

> Oups, sorry, I posted on the ML the patch as an attachment.
>
> Here is the pasteall link:
> http://www.pasteall.org/11157
>
> Damien
>
> Le 20 févr. 2010 à 15:35, Damien Plisson a écrit :
>
> > Hi Arystan,
> >
> > Looking at the float-to-string function, it appears it does bitwise
> operations on long type vars. And this particular 'long' type size changes
> when going from 32 to 64bit !
> >
> > Here is a patch that replaces the long type with fixed length int32_t. It
> makes the cube export work well :
> >
> >
> >
> > If you confirm this fixes the issue, I'll commit the updated darwin9 lib.
> >
> > Cheers,
> > Damien
> >
> > Le 20 févr. 2010 à 14:41, Arystan Dyussenov a écrit :
> >
> >> Damien, Ken,
> >>
> >> Zero-sized cube problem is fixed.
> >>
> >> This problem was caused by OpenCollada's own float-to-string conversion
> >> functions which don't work correctly on 64-bit systems.
> >>
> >> The fix was to use sprintf. The disadvantage of using sprintf is that
> >> precision has to be limited to N digits in fractional part.
> >>
> >> This patch modifies conversion functions to use sprintf with 6 digits
> after
> >> the decimal point stripping off trailing zeros.
> >>
> >> The patch is on the wiki:
> >> http://wiki.blender.org/index.php/File:Opencollada_ftoa_patch.diff
> >> Tested on 64-bit Mac OSX 10.6.
> >>
> >> The original float-to-string function looks like this:
> >> http://www.pasteall.org/11155/c
> >> It only works on 32-bit. On 64-bit, it at minimum ignores the sign. If
> >> anyone knows how to make it work on 64-bit, that'd be a real fix!
> >>
> >> Arystan
> >>
>
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>