Re: OT / OFF header info

Perhaps we could queue this up for discussion in the conference call
tomorrow?

As we had discussed earlier (when reviewing the known table tags and DSIG
table), it would be good to call out this table and the expectations around
it (not compatible with the pre-processing steps) in the WOFF 2.0
specification.

In reading through the WOFF 1.0 specification (specifically the NOTE in
Section 6. Font Data Tables), it would not appear as if enforcing digital
signatures was a MUST feature.  Rather, that WOFF 1.0 creation tools had
flexibility in modifying the font and dropping the table, and that
user-agents had flexibility in reconstituting the the font (in which case
the digital signatures would be invalid).

http://www.w3.org/TR/2012/REC-WOFF-20121213/

At the moment the WOFF 2.0 reference implementation has support for keeping
the DSIG table (enabled by default).  Given that the table would not be
valid with WOFF 2.0, it would be great to remove this option and always
remove the table.

As the DSIG table can be quite large (~5-6K), especially for subsetted
fonts, stripping the table will further reduce the web font files size.
 Yet another nice latency win!

On Tue, Apr 29, 2014 at 11:15 AM, John Hudson <tiro@tiro.com> wrote:

> On the subject of WOFF2 invalidation of OT digital signatures, I presume
> the only recommendation we can give to font developers who want digital
> signatures to be preserved in webfonts is to use WOFF1?
>
> Early discussions about webfont deployment included ideas around
> serialising font files and using custom tables behind the DSIG to record
> this. I don't know if any foundries are actually doing this, but in such a
> situation WOFF2 might be unattractive in this regard.
>
> JH
>
>
>