Re: [Freefont-bugs] a comparison of FreeMono.ttf versions of 2003-06-24 and 2008-03-24

[James Cloos]

>>>>> "Steve" == Steve White <address@hidden> writes:

Steve> I also tried the minspace option, and could never get it to do anything.

Steve> Can you point us to documentation?

I think the code snippit from libXft said it all.  With minspace=true
libXft uses accent+descent to determine the line height, otherwise it
uses freetype's metrics.height value.

Hmm.  I should've quoted a bit more:

,----[ excerpt from libXft/src/xftfreetype.c ]
| descent = -(face->size->metrics.descender >> 6);
| ascent = face->size->metrics.ascender >> 6;
| if (fi->minspace)
|     height = ascent + descent;
| else
|     height = face->size->metrics.height >> 6;
`----

So, then, the question is what freetype uses to define the members of
the metrics struct.

face is an FT_Face.  From freetype.h I see that the size member's
metrics member is an FT_Size_Metrics struct.

>From freetype.h itself:

 /*    ascender     :: The typographic ascender of the face,       */
 /*                    expressed in font units.  For font formats  */
 /*                    not having this information, it is set to   */
 /*                    `bbox.yMax'.  Only relevant for scalable    */
 /*                    formats.                                    */
 /*                                                                */
 /*    descender    :: The typographic descender of the face,      */
 /*                    expressed in font units.  For font formats  */
 /*                    not having this information, it is set to   */
 /*                    `bbox.yMin'.  Note that this field is       */
 /*                    usually negative.  Only relevant for        */
 /*                    scalable formats.                           */
 /*                                                                */
 /*    height       :: The height is the vertical distance         */
 /*                    between two consecutive baselines,          */
 /*                    expressed in font units.  It is always      */
 /*                    positive.  Only relevant for scalable       */
 /*                    formats.                                    */
 /*                                                                */

For a TTF font, the next thing to check is freetype2/src/sfnt/sfobjs.c
where there is this comment:

 /* XXX: Computing the ascender/descender/height is very different */
 /*      from what the specification tells you.  Apparently, we    */
 /*      must be careful because                                   */
 /*                                                                */
 /*      - not all fonts have an OS/2 table; in this case, we take */
 /*        the values in the horizontal header.  However, these    */
 /*        values very often are not reliable.                     */
 /*                                                                */
 /*      - otherwise, the correct typographic values are in the    */
 /*        sTypoAscender, sTypoDescender & sTypoLineGap fields.    */
 /*                                                                */
 /*        However, certain fonts have these fields set to 0.      */
 /*        Rather, they have usWinAscent & usWinDescent correctly  */
 /*        set (but with different values).                        */
 /*                                                                */
 /*      As an example, Arial Narrow is implemented through four   */
 /*      files ARIALN.TTF, ARIALNI.TTF, ARIALNB.TTF & ARIALNBI.TTF */
 /*                                                                */
 /*      Strangely, all fonts have the same values in their        */
 /*      sTypoXXX fields, except ARIALNB which sets them to 0.     */
 /*                                                                */
 /*      On the other hand, they all have different                */
 /*      usWinAscent/Descent values -- as a conclusion, the OS/2   */
 /*      table cannot be used to compute the text height reliably! */
 /*                                                                */

 /* The ascender/descender/height are computed from the OS/2 table */
 /* when found.  Otherwise, they're taken from the horizontal      */
 /* header.                                                        */
 /*                                                                */

and this chunk of code:

,----[ excerpt from freetype2/src/sfnt/sfobjs.c ]
|  root->ascender  = face->horizontal.Ascender;
|  root->descender = face->horizontal.Descender;
| 
|  root->height    = (FT_Short)( root->ascender - root->descender +
|                                face->horizontal.Line_Gap );
`----

So the difference in using minspace=true is that the Line_Gap isn't
included in the computation.  

AFAICT the current freetype code (I looked at cvs) uses the hhea table
for all three of those values.

-JimC
-- 
James Cloos <address@hidden>         OpenPGP: 1024D/ED7DAEA6