[cairo] Custom UserFont sizing questions

[Ian Britten]

Behdad Esfahbod wrote:

>> I'm down inside the 'render' function for my custom UserFont, and
>> am struggling with some size/scale issues.  I'm hoping someone can
>> help explain what I'm seeing, and how to best address+simplify it.

>> However, I now need to render my face using raw FT calls.

> Why exactly?

Ya, I figured I probably should have explained that, but the message
was getting pretty long...  :P

It came down to cap-styles.  I need to be able dynamically specify
whether the strokes are round or square for each font, and that is
not possible using my 'double-the-stroke-path-to-make-it-closed'
trick, since the ends actually become interior points of the path,
whose appearance is controlled by the join style (Always round),
not the cap style.

So, I'm back to an approach similar to our non-Cairo rendering,
where I use open paths for the font strokes, and some custom
callbacks to FT_Outline_Decompose() which handle dropping the last
closing point (Which was automatically added by FT).  With this, I
can correctly control the cap styles.

I know it's not necessarily 'right', but I'm kindof between a rock
and a hard place.  We're _legally_ required to produce correct
output (Both according to international Standards, and customer
contracts), and if I can't get everything drawn correctly with
Cairo, then we may not be able to use it (I'm still in a research
and evaluation phase).

[ Obviously, I'm open to suggestions/alternatives, but based on
discussions to date, this is the approach I've ended up with. ]



>> Basically, my (cairomm) render code looks like:
>>        FT_Set_Char_Size(face, 0, mmToPoint(115.238887) * 64,
>>                         72, 72);
>>        FT_Load_Glyph(face, glyph, load_flags);
>>        inContext->save();
>>        inContext->scale(0.001158/2.0, -0.001158/2.0);
   [ snip ]
>> Now,
>> - Can someone explain the 2.0?  I don't understand it...

Just a bit of an update:
Last night I found a stray '10' in my code from some earlier flailing
I was doing.  As such the '2.0' I mentioned seems to actually be
closer to '24.0' or so (Still flailing).  Everything else is unchanged.

I'm still looking into this, but am no closer to understanding where
that fudge-factor is coming from.  I'm guessing it's my code somehow,
but if anyone can describe/clarify what I _should_ be doing, or if
anyone sees anything suspect, I'd certainly appreciate it!

Many thanks for any help.
Ian