iiinova — where your digital self keeps evolving

By the way, the online tools have UTR #25 Math_Class: util.unicode.org/UnicodeJsps/...

000

I would expect to see U+2A75 in §4.3: util.unicode.org/UnicodeJsps/....

100

Doing no-precedence first makes it possible to come back to the precedence question separately.

000

In particular, if the language defines precedence, I would expect www.unicode.org/Public/math/..., which gives the « default » role of operators, to be taken into account, and the Rs (which by default have relation-like spacing in typesetting) to have precedence somewhere in the region of <=> to ==.

300

By « leaves room for defining precedence later if needed », I meant not just room for user-defined precedence (which has the problems you mention) but also room for language-defined precedence.

100

Right, that’s why you’re not adding a mechanism to define precedence. But similarly defining it yourself seems suboptimal: it’s trivial to come up with something that parses counterintuitively. Some cases would only compile one way unless you identify bool with ℤ/2ℤ, but some work both ways.

100

You might also find it interesting to skim www.unicode.org/reports/tr25..., in particular tables 9–11: operators can be sequences.

UTR #25: UNICODE SUPPORT FOR MATHEMATICS

For operators that « obviously » have a precedence lower than the arithmetic operators (set arithmetic, but also relations, etc.; it is pretty easy to write things that read differently than they parse) it would avoid misunderstandings—and it leaves room for defining precedence later if needed.

200

Robin Leroy @eggrobin.bsky.social · 20/08/2026

Have you considered going with « no precedence », rather than « fixed precedence », i.e., making 2*ℤ∪ℕ ill-formed, rather than equal to 2*(ℤ∪ℕ)? Ada does that for and vs. or and for ** with itself, see www.ada-auth.org/standards/2y.... For user-defined operators that approach seems useful.

100

The draft minutes of Unicode Technical Committee Meeting #188 (recorded by yours truly) have been posted: www.unicode.org/L2/L2026/261...

Luna @moonbase.lgbt · 17/08/2026

UTC 188 Minutes

011

Reposted by Robin Leroy

chat is it good when i have to change the browser zoom to 50% to fit github's entire incident report in a single screenshot?

screenshot of a firefox window at 50% zoom, showing just an absolute fuckload of github outages from today

0619

Reposted by Robin Leroy

Gro-Tsen @gro-tsen.bsky.social · 27/07/2026

So, IBM publishes machine-generated translations of AIX documentation on its Web site, and the French translation of awk documentation translated “sin” and “log” (the math functions) as if they were the words for a religious transgression and a journal respectively. 🤣

«
Fonctions arithmétiques

Les fonctions arithmétiques suivantes exécutent les mêmes actions que les sous-routines de langage C sous le même nom:

atan2( Y, X ) Renvoie l'arc tangente de Y/X.
Cos ( X ) Renvoie le cosinus de X; X est en radians.
Péché ( X ) Renvoie le péché de X; X est en radians.
Exp ( X ) Renvoie la fonction exponentielle de X.
Journal ( X ) Renvoie le logarithme naturel de X.
Sqrt ( X ) Renvoie la racine carrée de X.
»

814055

Robin Leroy @eggrobin.bsky.social · 30/06/2026

Note that most relevant operators are Pattern_Syntax: util.unicode.org/UnicodeJsps/... (Math_Class is defined in an informative report which lags far behind the standard and has no stability guarantees, Pattern_Syntax is immutable, so the UAX #31 definition uses the latter.)

util.unicode.org

Unicode Utilities: UnicodeSet Comparison

000

Robin Leroy @eggrobin.bsky.social · 30/06/2026

The idea is indeed to allow languages to define them as infix operators (without a need for backticking vel sim.), because we saw them used that way in languages that allow that (not so for the unary operators). There is some guidance for doing that in UAX #31, see www.unicode.org/reports/tr31....

100

Robin Leroy @eggrobin.bsky.social · 30/06/2026

Note that the characters normally used as infix operators (util.unicode.org/UnicodeJsps/...) have not been added to identifiers (and were never part of identifiers). What got added back was mostly characters that have letter-like/unary mathematical usage: util.unicode.org/UnicodeJsps/...

Robin Leroy @eggrobin.bsky.social · 17/06/2026

This reminds me I haven’t done Huehnergard exercises since… April 2025 I think? 😭 … Now I am considering doing some Akkadian exercises at half past two in the morning.

Implement Greek uppercasing by Manishearth · Pull Request #3629 · unicode-org/icu4x

En grec… il faut faire des choses horribles avec les accents, pour que άυλος devienne ΑΫΛΟΣ (icu.unicode.org/design/case/... ; j’ai travaillé sur une implémentation de la capitalisation du grec il y a trois ans, github.com/unicode-org/...).

github.com

This implements the Greek uppercasing behavior as sketched out in https://icu.unicode.org/design/case/greek-upper. The guts of the algorithm were written by @eggrobin, I added proper data. This use...

020

En irlandais, certaines lettres restent minuscules quand le texte est composé en capitales ; on ne va pas avoir un « h » différent du « h » pour que la transformation de « Parlaimint na hEorpa » en « PARLAIMINT NA hEORPA » soit non contextuelle.

Les mots « PARLEMENT EUROPÉEN » dans toutes les langues officielles de l’Union Européenne. En irlandais, « PARLAIMINT NA hEORPA ».

220

(Et de même qu’on applique parfois des règles orthographiques étrangères pour mettre un mot d’origine étrangère au pluriel, e.g., en anglais « lemmata », en allemand « Limites », « Volumina », les Turcs appliquent apparemment des règles étrangères pour mettre les noms étrangers en majuscules.)

Plus généralement, les transformations de casse ont l’air d’être des transformations de caractères simples, mais sont en fait des transformations orthographiques qui peuvent être complexes.

si je tape « Ankara », en lettres latines utilisées en français, je ne vais pas passer d’un clavier français à un clavier turc ; un utilisateur d’un clavier turc ne va pas passer à un clavier anglais pour taper « Biden » ni « BIDEN », mais utiliser le « i » et le « I » qu’il a sous la main.

En pratique, les gens qui tapent du texte ne sont pour la plupart pas des informaticiens qui veulent que le changement de casse soit une opération simple, et ce genre de désunification marche très mal :

Ça aurait été plus simple pour le changement de casse automatique, mais seulement dans un univers où les Turcs pensent que leur i latin est une lettre différente du i latin de tous les autres utilisateurs de l’alphabet latin. Et s’ils se mêlent les pinceaux entre les deux, ça aurait été l’enfer.

Robin Leroy @eggrobin.bsky.social · 28/05/2026

I used to work on some infrastructure that allowed—and strongly encouraged—software engineers to provide a screenshot of the string in the broader context of the UI. I once saw a screenshot of a button with a single word on it that had been carefully cropped to show only the button.

030

Reposted by Robin Leroy

Gro-Tsen @gro-tsen.bsky.social · 28/05/2026

A linguistic point about French vs English: I recently retrieved an item from an Amazon locker, and the locker thanked me, in French, with “merci d'utiliser Amazon locker”. I would argue this is a bad translation of “thank you for using Amazon locker”. Let me explain. 🧵⤵️ •1/17

2185

Robin Leroy @eggrobin.bsky.social · 10/05/2026

I’m not quite sure what’s baffling here, but then I am French and my background is pure maths…

030

Robin Leroy @eggrobin.bsky.social · 07/05/2026

C’est en tous cas documenté dans UTS #10 (www.unicode.org/reports/tr10... Table 5) et supporté par ICU : icu4c-demos.unicode.org/icu-bin/coll..., « backwards secondary ». Effectivement le Littré a cote < côte < coté < côté : archive.org/details/1883....

UTS #10: Unicode Collation Algorithm

020

Robin Leroy @eggrobin.bsky.social · 04/05/2026

The draft minutes of Unicode Technical Committee Meeting #187 (recorded by yours truly) have been posted: www.unicode.org/L2/L2026/260...

Robin Leroy @eggrobin.bsky.social · 03/05/2026

UTC 187 Minutes

050

It looks like I have been guilty of a couple of 60+ pagers in the past few years… They both have tables of contents, but “approachable” might not be the most fitting epithet for them… 😅 (www.unicode.org/L2/L2022/222... & www.unicode.org/L2/L2024/242...; I am nothing if not eclectic…)

Robin Leroy @eggrobin.bsky.social · 01/05/2026

La différence est claire ici :

Robin Leroy @eggrobin.bsky.social · 26/04/2026

Well, surely you would have gone to back to Italy anyway, so now that trip might be funded! Clever planning!

Robin Leroy @eggrobin.bsky.social · 26/04/2026

Seems like a good excuse to come back to Turin then?

Kate @katef.bsky.social · 18/04/2026

Reposted by Robin Leroy

the unicode.org ICU people used this as an example for DFA :3 unicode-org.atlassian.net/browse/ICU-2...

unicode-org.atlassian.net

Jira

1457

Reposted by Robin Leroy

John C. Baez @johncarlosbaez.bsky.social · 12/04/2026

Coxeter noticed that you can stick regular tetrahedra together to make a helix. It never repeats: no two tetrahedra have the same orientation in space! These simple discoveries are priceless. Click on the link to watch a short animation of it - it's fun! commons.wikimedia.org/wiki/File:Bo...

A helix formed by gluing together regular tetrahedra.

3495

Reposted by Robin Leroy

iximeow @there.is.no.aarch64.mov · 04/04/2026

120 second exposure and Orion is still so tiny

another greyscale image, there's a dot in the middle and stars are smears going towards the bottom right

1415

I can’t find scratch calculations lying around, but these 2014 maths lecture notes and these notes from an Assyriology conference last year are probably quite heinous too… (At the Assyriology conference my converter was broken so I was refilling every 15 min, though I don’t think that’s visible.)

A page of handwritten lecture notes from a course on the algebraic method in combinatorics, discussing the Borsuk–Ulam theorem.

A page of handwritten notes from an Assyriology conference, including the beginning of a talk by Devecchi and Chambon titled “Capacity units and administrative practice”

Right, I have noticed that it seems unusual/fancy to non-French Europeans too, whereas to me it’s just the only thing I write with (and I’m hardly serious about handwriting, most of this is ugly scribbles that I can barely read a few days later).

I am always surprised that people don’t use fountain pens at school, in France it was mandatory (and I’m 32), but apparently that’s not the case in other European countries. Certainly I very much dislike writing with a ballpoint compared to a fountain pen.

100

(For those who do not know Sumerian—but what kind of a scribe is that?—, see the translation etcsl.orinst.ox.ac.uk/cgi-bin/etcs...)

171

Robin Leroy @eggrobin.bsky.social · 16/03/2026

𒁾𒊬𒊑 𒈬 𒁹𒀀𒀭 𒃶𒂗𒍪 𒋗𒉌 𒃶𒊷𒊷 𒂊𒉈𒀀𒀭 𒁾𒊬𒊑

161

𒈪𒉆𒈠 𒀸 𒅆𒌓𒋾 𒀝𒅗𒁕𒀜𒌅 𒆷 𒋫𒆳𒋻𒈠 𒆷 𒌅𒈨𒁉𒆷 oracc.org/saao/P237990...

Robin Leroy @eggrobin.bsky.social · 15/03/2026

Reposted by Robin Leroy

Surely we also need more Sumerian content on Bluesky, so this seems as good a time as any to post this Blue Suede Shoes cover: www.youtube.com/watch?v=2gCe...

youtube.com

Blue Suede Shoes

YouTube video by Doctor Ammondt - Topic

021

Robin Leroy @eggrobin.bsky.social · 15/03/2026

Surely we also need more Sumerian content on Bluesky, so this seems as good a time as any to post this Blue Suede Shoes cover: www.youtube.com/watch?v=2gCe...

youtube.com

Blue Suede Shoes

YouTube video by Doctor Ammondt - Topic

021

Robin Leroy @eggrobin.bsky.social · 14/03/2026

seal.huji.ac.il/node/1736 (tab “textual fields”, col. ii ll. 7′ sqq.) and www.gkr.uni-leipzig.de/fileadmin/Fa... pp. 121 sq. The verb qatāru(m) mentioned by @moudhy.bsky.social is 𒋫𒂵𒀜𒌅𒌨 ta-qa₂-at-tu-ur ta-qattur-∅ 2-√qtr.IMPFV-2ms

seal.huji.ac.il

CT 46, 44 | Sources of Early Akkadian Literature

Robin Leroy @eggrobin.bsky.social · 14/03/2026

I wish I could say I typed this from reading the tablet, but my OB cursive is abysmal so this would have taken me ages; I read 𒀀𒈾 𒈠𒁕𒁴 (ana madātim, and even there I was helped by @moudhy.bsky.social’s translation “about many things”) and then looked it up, finding transliterations and translations…