Re: [ft-devel] otvalid modules

[Werner LEMBERG]

> I am working on a validator for the MATH table to slot it in to
> otvalid.

Thanks a lot!  I've added it to the CVS, with minor clean-ups and
formatting.

> I noticed that if the version number of the GDEF table is unknown it
> raises FT_INVALID_FORMAT, but if the version number of GSUB is
> unknown FT_INVALID_DATA is raised. Does this matter? I wasn't sure
> which to raise for the math table.

This is an inconsistency which I've just fixed in the CVS (and many
more too).  It should be FT_INVALID_FORMAT.

> I did not wish to change the API, but the OT validator currently has
> arguments in which all the other tables are returned. The result is
> that MATH is a second class citizen -- it gets validated, but not
> returned.

I suggest a new API, probably one which can be extended in the
(unlikely?) case the OpenType people define a new table.  Can you
provide something?

> This has been tested on a total of one font, because I've only got
> one font. It was also tested on FontForge's output on an additional
> 2 fonts, but I'm not sure that means anything (Since I wrote both I
> might have made the same mistakes both places -- I did take pains
> not to refer to my previous work when writing this, however).

Everything looks very good.  I just wonder why some `OTV_NAME_ENTER'
constructs are commented out.  Do you want to avoid a flood of tracing
messages?


    Werner