Re: [Devel] New cache sub-system implementation

[Detlef Würkner]

address@hidden (David Turner) wrote:

[...]
>     C - provide an "acquire/release" interface to each
[...]
>   See also the document at:
>  
>     http://www.freetype.org/freetype2/docs/cache.html
> 
>   that describes the cache sub-system design, how to use it,
>   and even how to create a few custom caches..

The document describes FTC_Image_Cache_Acquire(),
FTC_Image_Cache_Release(), FTC_SBits_Cache_Acquire() and
FTC_SBits_Cache_Release(). The first two are implemented, but not
defined in a header file, the last two are neither implemented nor
defined (in freetype2-current.tar.gz dated Oct-31).

Since it is not possible to lock the FT_Face and FT_Size objects
returned by the cache manager, I have a question:

Is the FT_Glyph returned from FTC_Image_Cache_Acquire() guaranteed
to keep valid until released with FTC_Image_Cache_Release() even when
the FT_Face where the FT_Glyph was created from may be freed by the cache
manager due to an intermediate call of FTC_Manager_Lookup_Face() for a
different FT_Face from a different application?

If yes, is it allowed to call FT_Glyph_Transform(), FT_Glyph_Get_CBox()
and eventually FT_Glyph_To_Bitmap() with the FT_Glyph as argument?

(Trying to implement a system-wide cache in a multitasking environment)

Ciao, Detlef
-- 
_ // address@hidden
\X/  Detlef Wuerkner, Langgoens/Germany