Re: status of hmtx transform proposal?

*tl;dr* I think we should add it to the spec, looks like it does indeed
save ~1%.

I ran a quick scan over the Google Fonts corpus using Cosimo's tool (
https://github.com/anthrotype/fonttools/tree/woff2_hmtx, thanks for that!)
to create two woff2's, one with hmtx optimization (if possible) and one
without so I could compare the post-compression impact of the change. I
saved just over 1% for fonts with savings, just under 1% overall. 77% of
the TTF files I tested saved bytes when hmtx optimization was enabled.

Bytes saved% size5171.1%For fonts that saved at least 1 byte3980.85%For all
fonts

Count%1465100.0%Fonts113377.3%Fonts that had gains (eg had all x-min == lsb)
33222.7%Fonts that had no gains

Cheers, Rod S.

On Tue, Jul 28, 2015 at 8:07 AM, Levantovsky, Vladimir <
Vladimir.Levantovsky@monotype.com> wrote:

>  Hi Cosimo,
>
>
>
> I believe the WG agreed to use two reserved bits as an additional
> transform version number for future extensibility. We also agreed that if
> we confirm the ‘hmtx’ transform to yield about 1% gains on average (as your
> data shown) we will go ahead and implement it into the spec. This is
> something we plan to discuss tomorrow during the WG telcon (
> https://www.w3.org/Fonts/WG/track/actions/open, action 176).
>
>
>
> Thank you,
>
> Vlad
>
>
>
>
>
> *From:* Cosimo Lupo [mailto:cosimo.lupo@daltonmaag.com]
> *Sent:* Tuesday, July 28, 2015 5:02 AM
> *To:* public-webfonts-wg@w3.org
> *Subject:* status of hmtx transform proposal?
>
>
>
> Hello,
>
>
>
> I just wanted to know what is the situation of the hmtx transform that was
> proposed a few months ago.
>
>
>
> I remember Jonathan Kew proposed to use some of the reserved bits in the
> WOFF2 directory entries to specify a version number that would indicate the
> kind of transform to apply for each table.
>
>
>
> Are you planning to proceed along these lines and add this to the WOFF2
> spec?
>
>
>
> Thank you,
>
>
>
> All best,
>
>
>
> —
> Cosimo Lupo
>