Re: [ft-devel] is that a bug in freetype on allowing INSTCTRL to be called from glyphs?

[Nikolaus Waxweiler]

> There are two blocks of cryptic and even sarcastic(?) comments below
> that from both Infinality and Nikolaus Waxweiler about subpixel
> hinting which seems to suggest clear type fonts violates that
> intentionally, if I understand the comments correctly? (I think the
> sarcastic(?) comment is left by Nikolaus Waxweiler?). Therefore
> FreeType possibly does not throw an error because there are fonts
> which do that, quite intentionally; and in any case, FreeType ltself
> probably does not need to care whether  INSTCTRL is being called from
> glyphs or cvt.

It wasn't meant to be sarcastic,
https://www.microsoft.com/typography/cleartype/truetypecleartype.aspx#Toc227035738
says that

> Although the other selector flags associated with INSTCTRL are
> limited to the CVT Program, selector three can be used in either the
> CVT Program or the glyph program[10]. [10] This is contrary to the
> current OpenType specification but intended in the design of this
> selector. To make sure that one glyph does not change the settings
> for another glyph, if the instruction selector is set at the
> beginning of a glyph, it should be cleared at the end of the glyph.
> (This is not required if the instruction selector is used exclusively
> in the CVT program.)

I think I made it so that the flag is reset to the global CVT state
after each glyph so that no side effects can occur. I hope.