Re: [ft] Own stream implementation

[mpsuzuki]

Hi,

On Wed, 22 Sep 2010 23:39:03 +0200
Christoph Nelles <address@hidden> wrote:
>i am currently trying to link PhysicsFS as file system abstraction layer
>with Freetype2.  So for own implemented stream, even after reading the
>documentation and looking into the code i still don't know who is
>responsible for deleting the FT_StreamRec when FT_Done_Face() is called.

Basically, FT_StreamRec is freed by FT_Stream_Free()
function. FT_Done_Face() invokes destroy_face(), and
destroy_face() invokes FT_Stream_Free(). So, most
FT2 clients don't invoke FT_Stream_Free() by themselves.

>Will the function free the record after closing the stream, or is up to
>to the caller to keep a reference and free it up himself after calling
>the function? 
>Is there anything to regard? Has anybody perhaps some public examples
>for custom streams, haven't found anything about this on google.

Yes, rarely we receive the request of documentation for
yet another stream object. Maybe it's because most people
make the memory image of the font from their special
storage and open a FT_Face object from the memory image,
and they are less motivated to implement their own stream
object.

# When I worked for "partial stream" 2 years ago,
# I referred gzip/lzw streams.

Regards,
mpsuzuki