Unwanted vulgar fractions

[laughinghan]

For some reason, vulgar fractions like 1/2 and 1/3 are also ending up in Symbola-basic, even though we aren't asking for them.

[laughinghan]

Conjecture: Latin-1 (specifically, ISO 8859-1) is included in its entirety by default, rather than just ASCII?

[jwmerrill]

Glyphs we ask for:

̱!()*+,-./0123456789:;<=>˂˃[]{|}~μ÷⟨⟩π–—∂∏∑−√∫≈≤≥

Glyphs we end up with:

[jwmerrill]

So it's not like we're getting all of latin 1. It's really just the vulgar fractions. And space for some reason.

[laughinghan]

Bizarre

Han

On Fri, Jun 3, 2016 at 11:40 AM, Jason Merrill notifications@github.com
wrote:

So it's not like we're getting all of latin 1. It's really just the vulgar
fractions. And space for some reason.

—
You are receiving this because you authored the thread.
Reply to this email directly, view it on GitHub
#1 (comment),
or mute the thread
https://github.com/notifications/unsubscribe/AANyEVmMv3Nu6r9xbYhg7mZqcgua4Cu7ks5qIHUSgaJpZM4ItuMv
.

[jwmerrill]

I just took a look at using the pyftsubset tool that's part of fonttools (which can be installed via HomeBrew on OS X), instead of the random perl library that's currently being used. The good news is that pyftsubset doesn't include vulgar fractions for no reason. The bad news is that its output crashes Chrome 51!

It has a lot of options, but I tried a bunch of the obvious ones, and I can't find anything that keeps the output from crashing Chrome.