Re: [ft] Strange hinting of cff font

[Werner LEMBERG]

> BTW in this source code we have also one small change to comment
> these lines:
>
> -        if ( fitted < scaled )
> -          x_scale -= x_scale / 50;
> +       /* if ( fitted < scaled )
> +          x_scale -= x_scale / 50;*/
>
> It does also better results as in some fonts and scales this does
> not preserve aspect ratio and is really strange constant anyway.

This is a heuristical constant; similar code is in the autohinter
(function `af_latin_metrics_scale_dim').  I'm quite sure that there
are reasons for it (unfortunately, there's no comment which explains
the adjustment sufficiently) -- maybe it's necessary to `sharpen' the
constraints to apply this miniscule scaling.

Can you provide samples where this scaling is harmful?  Or do you
observe a bad effect in general?


    Werner