[ft-devel] Re: Some public APIs using FT_Int/FT_UInt should be improved in future

[Werner LEMBERG]

> I'm sorry, you told to use FT_Long, but I committed FT_UInt32.
> 
>    104    FT_EXPORT_DEF( FT_Error )
>    105    FT_Get_Advances( FT_Face    face,
>    106                     FT_UInt    start,
>    107                     FT_UInt    count,
>    108                     FT_UInt32  flags,
>    109                     FT_Fixed  *padvances )
>    110    {
> 
>       ...
> 
>    143      flags |= FT_LOAD_ADVANCE_ONLY;
>    144      for ( nn = 0; nn < count; nn++ )
>    145      {
>    146        error = FT_Load_Glyph( face, start + nn, flags );
> 
>       ...
> 
> As you can see, FT_Get_Advances() passes its 4th argument flags to
> FT_Load_Glyph() whose 3rd argument is declared as FT_UInt32 in
> public interface.

Actually, it's FT_Int32 (for reasons I no longer know):

  FT_EXPORT( FT_Error )
  FT_Load_Glyph( FT_Face   face,
                 FT_UInt   glyph_index,
                 FT_Int32  load_flags );

> To match their types, I declared the flag as FT_UInt32.

Hmm.  Signed vs. unsigned...

> BTW, if there's any criterion to determine "FT_ULong" is better
> in public API (or "FT_UInt32" is better), please let me know.

For me, FT_Long and FT_Int look more `natural' than, say, FT_Int32.
But this is a feeling only.


    Werner