HarfBuzz 10.0.0 released

[Posted September 24, 2024 by jzb]

Version 10.0.0 of the HarfBuzz text-shaping engine has been released. Notable changes in this release include Unicode 16.0.0 support, adding Cairo script as an output format for hb-view, and a number of bug fixes.

to post comments

In case you want to know

Posted Sep 24, 2024 21:18 UTC (Tue) by python (guest, #171317) [Link] (7 responses)

"Why is it called HarfBuzz?

HarfBuzz began its life as text-shaping code within the FreeType project (and you will see references to the FreeType authors within the source code copyright declarations), but was then extracted out to its own project. This project is maintained by Behdad Esfahbod, who named it HarfBuzz. Originally, it was a shaping engine for OpenType fonts—"HarfBuzz" is the Persian for "open type". "

I think it is how the Persian نوع باز would be pronounced in English

I didn't know I wanted to know

Posted Sep 25, 2024 6:29 UTC (Wed) by cbushey (guest, #142134) [Link]

I found that remark very interesting and I'm glad you took the time to inform the reader base. Pretty much +1 but with actual comment.

In case you want to know

Posted Sep 25, 2024 8:43 UTC (Wed) by a3f (subscriber, #106306) [Link] (4 responses)

Small correction: it's حرف باز. The word نوع means kind/category/type, while حرف means letter.

(Thanks for your comment btw, I didn't know that Buzz was Farsi for Open, Harf happens to be the same word in Arabic as well.)

In case you want to know

Posted Sep 26, 2024 5:39 UTC (Thu) by milesrout (subscriber, #126894) [Link] (3 responses)

Apparently it is a loanword from Arabic to Persian.

(Why do we call them loanwords or borrowings, does anyone know? It isn't like they're ever given back. It is more like word piracy if you ask me. And English is the biggest word pirate of them all.)

In case you want to know

Posted Sep 26, 2024 8:05 UTC (Thu) by tzafrir (subscriber, #11501) [Link]

Really off topic, but sometime a word is given back. For instance, anime.

In case you want to know

Posted Sep 26, 2024 8:07 UTC (Thu) by Wol (subscriber, #4433) [Link]

> And English is the biggest word pirate of them all.)

Fancy a mug o' char?

Cheers,
Wol

Loan words, was: In case you want to know

Posted Sep 27, 2024 18:21 UTC (Fri) by MortenSickel (subscriber, #3238) [Link]

In Norwegian we have taken into use quite a few english words, but looking into them, many have old Norse roots, so it is really kind of taking back.

Glad I finally know

Posted Sep 26, 2024 20:35 UTC (Thu) by hkoosha (subscriber, #154794) [Link]

For a decade, I didn't know the name means Open Type/Letter, as the buzz (or baz) suffix in Persian is _almost_ same as the "-er" suffix in English, e.g. player, or gambler; the one who gambles.

So HarfBuzz for me was the program that plays with letters, likes letters, manipulates them, etc. Harf also means speak or speach, so for a while I read it as something which likes to speak a lot. In my defense, "Open Type" in Persian should probably be written as "Buzz Harf" or maybe "Harf-e-Buzz" (the order or the 'e' in-between does some heavy lifting).

Funny how I speculated the name so many years yet never looked it up properly.

Harfbuzz and TeX

Posted Oct 3, 2024 9:25 UTC (Thu) by callegar (guest, #16148) [Link]

Compliments for the release, and a question in case some expert of Harfbuzz might be able to answer.

Among the huge range of software internally using Harfbuzz, a niche but notable subset is certainly given by recent TeX engines (LuaTeX and XeTeX). TeX is notable in that it puts an extreme emphasis on the quality of the layout, wrt most other software (e.g., word processors). For instance, it is renown for having one of the best line breaking algorithms and for offering facilities for microtypography (e.g., https://ctan.org/pkg/microtype). This clearly stretches the requirements posed on the font renderer.

Currently, XeTeX uses Harfbuzz only, and while this choice is generally fine, there are a few corner cases (https://github.com/latex3/luaotfload/issues/152#issuecomm...). On the other way round, LuaTeX historically had its own text renderer ("node") and more recently switched to also offering a renderer based on HarfBuzz as an alternative. Working with node, LuaTex ends up being way slower than XeTeX. The HarfBuzz renderer makes LuaTeX significantly faster at handling fonts (and because this task is a "slow phase" makes LuaTeX faster as a whole). Yet, it brings in significant defects (including a tendency to miss good hyphenation points). An attempt at trying to document the pros and cons of each renderer is in https://github.com/latex3/luaotfload/issues/265.

Because of the speed it would be great if LuaTeX users could switch to the Harfbuzz renderer most of the time, but apparently as of today there are some unavoidable trade-offs and when using Harfbuzz, XeTeX and LuaTeX also end up dealing with them differently (which is justified by LuaTeX having an alternative to Harfbuzz). A thorough discussion is available in https://github.com/latex3/luaotfload/issues/152.

To the best of my understanding, the main issue is that "TeX model of discretionary line break is rather unique so neither HarfBuzz or any other text layout library I’m familiar with has any special support for the way TeX does it", which has consequences, introducing cases that basically cannot be handled " If you need both stacking diacritics and proper hyphenation you have a problem".

What I wonder if Harfbuzz might have on the drawing board some ideas or plans for interaction with the developers of the modern TeX engines and for providing a better support to TeX, which might eventually end up in better typography in other fields as well.