Isn't it just +4 bytes for the UInt32 with the tag value? ( http://dev.w3.org/webfonts/WOFF2/spec/#table_dir_format) On Fri, May 29, 2015 at 10:18 AM, Behdad Esfahbod <behdad@google.com> wrote: > On Fri, May 29, 2015 at 10:14 AM, David Kuettel <kuettel@google.com> > wrote: > >> On Thu, May 28, 2015 at 7:24 PM, Levantovsky, Vladimir < >> Vladimir.Levantovsky@monotype.com> wrote: >> >>> We can definitely revisit the list of known tables and see if anything >>> there could be eliminated / swapped for 'meta' tag, but for all intent and >>> purposes the list has nothing to do with how compression is applied, it >>> only helps to save some bytes by reducing the size of the table directory. >>> >>> Any arbitrary table that is not "known" would be represented in the >>> table directory by the 'arbitrary' flag followed by the table tag, the >>> table data itself will be a part of the compressed data stream regardless >>> of whether it's known or not. >>> >> >> It would be great to actually measure this for a font with a 'meta' >> table, just to have the hard data. Any volunteers? >> > > I believe we know the difference to be < 4 bytes, no? > > >> >>> Thank you, >>> Vlad >>> >>> >>> On May 28, 2015, at 7:29 PM, "Sergey Malkin" <sergeym@microsoft.com> >>> wrote: >>> >>> Apple defines ‘meta’ table that is still outside of OpenType spec. It >>> contains information about languages font supports and designed for. We in >>> Windows 10 decided to add this table to our system fonts and are in process >>> of doing so. >>> >>> >>> >>> I noticed that this table is not in the list of known table tags ( >>> http://www.w3.org/TR/WOFF2/#table_dir_format) and this may be the only >>> real world table that is not compressed by this mechanism. Can we consider >>> adding this table to the list? I know that list is currently full, but >>> maybe we can replace some rarely used table with it? >>> >>> >>> >>> Thanks, >>> >>> Sergey >>> >>> >>> >>> >>> >>> >> >
Received on Friday, 29 May 2015 17:47:55 UTC