Re: [Freetype] API to tell if TrueType hinter compiled in?

[David Turner]

Hi Antoine and all,

I'd like to implement all of this, but my main priority is
getting 2.0.6 out of the door as soon as possible..

After that, I envision a few major changes within FreeType,
and these would probably include a version/features interface

Of course, if anyone has patches to contribute, we'll consider
them for inclusion if you provide them now..

More on this later..

Cheers,

- David

> 
> Other elements that may be of interest include:
>
> - version numbers, as signaled by Brian;
> - gray-scaling support;
> - autohinting;
> - max number of loadable extensions (already available at runtime?);
> - threading-safety;
> - cache support (and where);
> - DLL/so/static build (is it possible?);
> - with limited support for big/CJKV fonts (of concern for 16-bit
>     implementations only);
> and all the options listed in freetype/config/ftoption.h:
> - convenience functions;
> - alternate glyph image support;
> - psnames;
> - Adobe AGL support;
> - embeeded bitmaps support;
> - access to sfnt 'name' table;
> - T1 hinter;
> - T1 AFM support;
> - MM support;
> etc. (more?)
> 
> The available schemes range from:
> 
>   FT_EXPORT( FT_Error )
>   FT_Get_Capabilities( FT_some_fixed_structure * );
> 
> which is not very extensible, includes
> 
>   FT_EXPORT( FT_Error )
>   FT_Get_Capabilities( unsigned  minimum_version_requested,
>                        FT_some_structure_depending_of_version * );
> 
> which is better but not perfect, or
> 
>   FT_EXPORT( FT_Error )
>   FT_Get_Capabilities( unsigned  index,
>                        void * );
> 
>   /* the infos returned in the structure depends of the info requested */
> 
> which is quite heavy (big switch sparsely used) and deserves a
> bad perfs/price ratio, to
> 
>   FT_EXPORT( FT_Error )
>   FT_Get_Capabilities( unsigned  size_struct_passed,
>                        void *infos );
> 
>   /* the library fills only the `size_struct_passed' first bytes */
> 
>   /* User code uses a struct defined statically depending of the .h
>      used hence the version of library the source were compiled for.
>      Call sequence is
> 
>      struct FT_Capabilities caps;
> 
>      //...
> 
>      if( FT_Get_Capabilities( sizeof caps, caps ) ...
> 
>    */
> 
> 
> OTOH, if we are on the way to extend the API, I have another
> suggestion: with the TrueType rasterizer, we always masquerade
> as the Windows 3.1 engine (in GETVERSION[]); since the newer
> engines have additional features like gray-scaling that some fonts
> may decide to use/don't, it appears to me it would be useful to
> return a customizable (by user app) value to GETVERSION[].
> Perhaps there are similar "magic" values in the other modules, too
> (size of cache comes to mind).
> 
> Antoine
> 
> _______________________________________________
> Freetype mailing list
> address@hidden
> http://www.freetype.org/mailman/listinfo/freetype