Re: [ft-devel] The new v40 interpreter

[Hin-Tak Leung]

--------------------------------------------
On Tue, 2/8/16, Werner LEMBERG <address@hidden> wrote:

> > I wondered whether I should hook up v38 instead, or in addition.
> > Speed is definitely not a concern for a tool that uses private
> > version of FreeType to analyze fonts - it will just take whatever
 > reasonable time it takes to accomplish the task...
 
> I think v40 is a better choice for validation.

For (my own?) goals of validating fonts for usage on libre platforms and 
'quality in general', whatever is FreeType's default needs to be offered.
Since v40 is to be default on, that's that. OTOH, there are a whole bunch of 
users (font designers, of the probably commercial type, or at least
hoping-to-make-money-from-font-design type) who views it as a check for 
'fitness to run on MS windows' - at least they looked at the old font validator 
that way.

That was the first response on typedrawers on my announcement of the beginning 
of the new test: how can one use one rasterer checks for errors targeting 
another?

In reality, most errors with hinting instructions (I see on my ~5000 fonts) are 
not rasterization errors, but programming errors - mis-counting the number of 
points/contours, running the operand stack dry, etc. These are errors on *any* 
platform.

For those errors which are rasterization errors - i.e. actual pixels being 
drawn/not-drawn the "right" way - would v38 be closer to how MS windows does 
things?