rosettatype / yrsa-rasa

Yrsa & Rasa: fonts for Latin and Gujarati

Yrsa & Rasa

Yrsa and Rasa are open-source type families published by Rosetta with generous financial support from Google. The fonts support over 331 languages in the Latin script and 3 languages in the Gujarati script (Gujarati, Kachchi, Avestan). The family currently has 5 weights. The design was done by Anna Giedryś (@ancymonic) and David Březina (@MrBrezina). The font engineering was done by Johannes Neumeier (@kontur).

Yrsa is the name of the Latin-only type family. Rasa is the name of the Gujarati type family. Both type families are intended for continuous reading on the web (longer articles in online news, magazines, blogs). In Yrsa, a special consideration was given to Central and East European languages and proper shaping of their accents. Rasa supports a wide array of basic and compound syllables used in Gujarati. In terms of glyphs included Rasa is a superset of Yrsa, it includes the complete Latin.

What makes the Yrsa & Rasa project different is the design approach. It is a deliberate experiment in remixing existing typefaces to produce a new one. The Latin part is based on Merriweather by Eben Sorkin. The Gujarati is based on David Březina’s Skolar Gujarati.

The project has been documented in several articles. See the Yrsa & Rasa project page for details.

In 2020–2021, Yrsa and Rasa received a major update. Rosetta added a substantial set of glyphs to the uprights of both type families to provide wider language support, including full support for Vietnamese, additional stylistic sets of numerals (there are lining and text numerals, superiors, inferiors, fractions) and quite a few monetary symbols. On top of all this, Anna Giedryś created the highly anticipated and often requested italics for the Latin script. Unlike the uprights, the italics were designed completely from scratch, giving the family a distinctive touch. Optimized for screen and online use, Yrsa is now a well-rounded text typeface. Rasa received only a minor maintenance update and an addition of glyphs to support the Avestan language. Both families come with variable fonts!

Download

You can download the compiled fonts from Releases section of this repo or from the Google Fonts directory (soon). Alternatively, you can find them in Adobe Typekit.

Building the fonts from source

If you want to compile the fonts from sources, read the production README.

Rosetta – world typography specialists

Rosetta addresses the needs of global typography. Together with our collaborators we create original fonts for a polyphonic world. Our work has been featured and awarded. But most importantly it has enabled people to read better in their native language.

So far our retail library supports pan-European Latin, Arabic, Armenian, Greek, Cyrillic (besides Slavic languages we also support many Asian languages), Inuktitut, and Indic scripts like Gujarati and Devanagari. In total, the library covers more than 450 languages.

For more information see the website at rosettatype.com.

You can contact us at [email protected].

License

The fonts and related code are licensed under Open Font License. See OFL.txt for licensing information.

Feedback & progress

This version is now considered final. Let us know if you spot any problem (via issues).

yrsa-rasa's Issues

Yrsa version 2.004 missing from assets

Yrsa v2.004 version seems to be missing from assets. Instead the 2.004 version of Rasa seems to be listed twice.

Gujarati Ja + Nukta causes collisions / cannot be combined with matra i

Reported by Ivo Michl over email:

pri kombinaci JA+NUKTA se nukta stale zobrazuje dole a tim padem je kolizni,
pozadovana kombinace znaku dodana na pozici 0AF8 vypada spravne, nespojuje se vsak s dalsimi kombinujicimi znaky (samohlasky, virama...), totez plati pro ZHA...
napr. "J.I" by melo byt zapsano jako JA+NUKTA+I, coz se mi momentalne nedari, ale je mozne ze delam neco spatne... (?)
castecne to resi nukta na pozici 0ACD, tak lze zapsat JA s nuktou na spravnem miste a kombinovat s viramou, s ostatnimi znaky (I) vsak jiz spravne kombinovat nelze...

Lining "9" is off the baseline

The lining "9" is above or below the baseline at some sizes.

https://github.com/rosettatype/yrsa-rasa/blob/1f2caa8a5eb98ba087963e0cbf25c13b5816687b/fonts/Yrsa/Yrsa-Regular.ttf in Chrome 95 on Win10:

Copyright notice should match files and LICENSE.txt

LICENSE.txt is different to the font metadata, which is better:

Copyright (c) 2015 by Rosetta Type Foundry s.r.o. ([email protected]).

Copyright string must be ascii

Your copyright strings are non-ascii, eg

    <string>Copyright © 2016 by Rosetta Type Foundry. All rights reserved.</string>

https://github.com/rosettatype/yrsa/blob/master/Production/UFOs/Rasa-Light.ufo/fontinfo.plist#L10

I suggest that, similar to

Copyright (c) 2014, Vaibhav Singh (design) and Rosetta Type Foundry s.r.o. (post-production).

https://github.com/rosettatype/eczar/blob/master/Design/Eczar-Bold.glyphs#L37

We change it to

    <string>Copyright (c) 20XX Rosetta Type Foundry ([email protected])</string>

where the year is the first year you started working on it.

Yrsa: 'frac' with wrong numbers

It seems there's a problem with OTF 'frac' feature (Yrsa version 2.002) as numbers are modified (e.g. 1234567890 → 2345678901).

Example in LibreOffice:

• 12345/12345 with 'Yrsa' font
• the same text using 'Yrsa:frac'

Add interrobang!?

Gujarati half forms do not form correctly in MS Word?

There is a problem in displaying half characters in Gujarati of the Rasa font in Windows text editors such as MS word, Wordpad etc... The problem is that Rasa font is not displaying half characters correctly. Instead of joining the half letter to the letter next to it, MS Word (like all text editors in windows) shows it as a 'halant' which in most words considered as a bad or incorrect gujarati. However on Google Fonts it is displayed correctly. I have attached a screenshot for your better understanding.

This may be MS Word refusing OpenType layout or using older layout (gujr, but not gujr2).

(reported by Vaibhav Pandya over email)

Dutch "ij" rendered incorrectly on Firefox in Linux

The Dutch "ij" only gets one letter worth of space, but takes up two letters worth of space. The result is that the next character slightly overlaps the preceding "ij".

For example, several occurences in the lead of this article: https://www.hpdetijd.nl/2018-08-23/onduidelijkheid-han-ten-broeke/

All Dutch-language websites using the Yrsa font seem to show this problem, leading me to believe it's a problem with Yrsa

Yrsa webfonts: German ß sometimes renders as f

It seems - tested with the standard webfonts - creates an unwanted ft ligature if an t is followed.

Example: If you write heißt it displays incorrectly as heift. To avoid this you have to use font-variant-ligatures: no-common-ligatures in CSS to disable ligatures.

Internal font naming not canonical

Semibold should be SemiBold in the filename and internal metadata. This is annoying to fix on Mac OS X because it has a case insensitive filesystem.

Also the style names are not canonical. This spreadsheet details what I consider canonical.

I'm also updating https://github.com/googlefonts/gf-docs/blob/master/ProjectChecklist.md together for reference on this stuff. Should have done that a long time ago!

Here is the report from the checking tools:

$ mkdir -p ~/src/github.com/rosettatype/; cd !$;
$ git clone https://github.com/rosettatype/yrsa.git;
$ mkdir ~/src; cd !$;
$ git clone https://github.com/google/fonts.git; 
$ mkdir -p ~/src/github.com/google/fonts/ofl/rasa; cd !$;
$ cp ~/src/github.com/rosettatype/yrsa/Fonts/Rasa*ttf .;
$ cp ~/src/github.com/rosettatype/yrsa/LICENSE.txt .;
$ cp ~/src/github.com/rosettatype/yrsa/README.md ./DESCRIPTION.en_us.html;
$ for i in `ls -1 *ttf`; do echo $i; ot-sanitise $i; done; # passes! good :)
$ cd ~/src/github.com/google/fonts/tools;
$ python add_font.py ../ofl/rasa;
$ python sanity_check.py ../ofl/rasa;
FAIL: Rasa METADATA normal/600 postScriptName expected Rasa-SemiBold, got Rasa-Semibold (../ofl/rasa)
FAIL: Rasa METADATA normal/600 fullName expected Rasa SemiBold, got Rasa Semibold (../ofl/rasa)
FAIL: Rasa normal/300 'name' family[1] expected Rasa, got Rasa Lt (../ofl/rasa)
FAIL: Rasa normal/300 'name' fullName[1] expected Rasa Light, got Rasa Lt (../ofl/rasa)
FAIL: Rasa normal/500 'name' family[1] expected Rasa, got Rasa Me (../ofl/rasa)
FAIL: Rasa normal/500 'name' fullName[1] expected Rasa Medium, got Rasa Me (../ofl/rasa)
FAIL: Rasa normal/600 'name' family[1] expected Rasa, got Rasa Sb (../ofl/rasa)
FAIL: Rasa normal/600 'name' postScriptName[0] expected Rasa-SemiBold, got Rasa-Semibold (../ofl/rasa)
FAIL: Rasa normal/600 'name' postScriptName[1] expected Rasa-SemiBold, got Rasa-Semibold (../ofl/rasa)
FAIL: Rasa normal/600 'name' fullName[0] expected Rasa SemiBold, got Rasa Semibold (../ofl/rasa)
FAIL: Rasa normal/600 'name' fullName[1] expected Rasa SemiBold, got Rasa Sb (../ofl/rasa)

Add gujarati letter ZHA <U+0AF9>

https://unicode.org/L2/L2013/13143-gujarati-zha.pdf

Excessive space inside "Hi" with kerning on

https://github.com/rosettatype/yrsa-rasa/blob/eaa91e7eb3dcc87a05ebeb9fe6ff0a7c8d28e39d/fonts/Yrsa/Yrsa-Regular.otf at 60 ppem in Chrome 89 on Win10, with kerning enabled:

Same with kerning disabled:

ETA on Google Fonts approval?

Hey. I'm currently using Google Fonts to serve fonts in a few large scale applications and would rather avoid having to serve up Rasa via @font-face if I can help it.

As per title, do you have an ETA on Google Fonts adding Rasa et al to their fonts repo?

Thanks.

Have you considered adding oldstyle figures?

Hello Rosetta and contibutors,
I suppose that lack of oldstyle figures in yrsa-rasa is deliberate, but I think incorporating them would enrich the design greatly, and make it look even more beautiful and have more humanistic appearence in running texts containing dates or other arabic numbers.
I'm sorry I'm not a type designer, and I'm just lacking basic skills to contribute to your project.
That is why I'm asking, if according to you that would make sense to include this set of ten glyphs, to make the design complete, Having OSF seams to me quite basic.
What do you think?
Martyn

TTF files are broken

They have been removed for now. Once we figure out the cause of the issue (using and rendering of some glyphs on Windows breaks things) we will put them back.

Yrsa 'i' not rendered in MSIE 11

• https://twitter.com/PrestonStahley/status/906223732661911553