Bidi Brackets for Dummies

https://www.unicode.org/notes/tn39/

by bkudria • 6 years ago

32 27 6 years ago

Just like the Hindu-Arabic number system enabled great innovation in mathematics because it was so much easier to use than the Roman system, I wonder if the simplicity of characters in English, enabled the US to jump to such an early start in computer software.

English characters are pretty compact (26 symbols), no accent marks that can't be ignored, have 1:1 mapping between uppercase and lowercase, and are easy to break up by word. This enables even very simple algorithms in a very resource constrained computer to do some work that is mostly right.

For example, splitting a person's name by space, taking the last word, making it uppercase and sorting by ASCII value lexicographically, mostly works if you want to produce a phone book listing, especially in 1950s-1980's America. And you can code this with very simple integer operations without needing a lookup table or a bunch of special rules. Soundex is also pretty simple to implement and deals with a lot of homophones.

Of course, now that we have the computing resources and libraries, handling the vast diversity of human languages is doable, but in terms of bootstrapping a computer software industry, simplicity, I think played a role.

ouid• 6 years ago

Simplicity of keyboard seems like a bigger deal.

cryptonector2• 6 years ago

I suspect the simplicity of the English (Latin) alphabet was only of marginal importance in the U.S. having such a long lead in computer science and engineering over the rest of the world. Need (WWII, Cold War), population, industry (the U.S. produced 50% of the world's GDP at the end of WWII), wealth, a market economy, and all that goes with and precedes all of the preceding list (education, opportunities, institutions, culture, etc.) meant that the U.S. was bound to get started on computer science very early, and was bound to make great progress.

If English had had twice the number of glyphs, IBM would still have existed, even though they'd have had more work to do. There might have been some number of glyphs past which the burden might have slowed progress enough that some other nation would have been competitive, but it's all just information, and ultimately computer engineers would have figured out something.

Or look at some evidence: Japan already had what it needed to communicate digitally before WWII in spite of having a complex script (three, four if you include romaji), and most computer engineering and computer science progress came after WWII, and indeed, Japan's computer industry did reasonably well post-WWII -- comparable to France's, which did quite well considering how much smaller France was than the U.S. in population, industry, wealth, etc. And: need was decidedly critical in the UK's computer engineering and science development during WWII (Alan Turing, Colossus). France, Japan, and the UK, each had some of the conditions that the U.S. also had, but not as many, so it's not surprising that the U.S. did well.

I've a feeling that among the things that helped the U.S. that I didn't list above are also: the decadal census (which is intimately related to IBM's rise), Sears & Roebuck (the antecedent to Amazon), the immense geography of the country and widely dispersed population (which impacted the preceding two items). I bet others can add items to the list. Some of these certainly existed elsewhere.

Of somewhat less importance is that a number of very important mathematicians who helped found computer science (e.g., Claude Shannon, Alonzo Church, Haskell Curry) were Americans. Their research was published, of course, and many others were not Americans (e.g., Charles Babbage) which is why I say they were of less importance.

WorldMaker• 6 years ago

And: need was decidedly critical in the UK's computer engineering and science development during WWII (Alan Turing, Colossus)

It's useful to note here too that UK's work in WWII was started by desperate Polish need, and there is indication they got somewhat further ahead than people tend to credit them for. (It's interesting to wonder if Poland may have been ahead in computation if it wasn't, you know, invaded, and forced to kick the ball over to the UK.)

It's also worth noting the different approaches to "war time secrets" the US and the UK took. Where Alan Turing had a strong practical head start on computation in WWII, his work was locked under confidential and secret designations and he was not allowed to commercialize it after the war. Worse, he was almost entirely stopped from building practical machines, and it's a wonder he managed to contribute as much to computing theory as he did even with the restrictions he was under. (Also don't forget his own government lead him to an early suicide.)

The US in its contractor-based approach (including IBM's involvement) didn't hamstring commercial interests anywhere near as strongly (despite declassified hindsight now telling us they were straggling behind, comparatively, UK's efforts) so much after WWII by locking things into a "classified vault" and the US did not try stopping the people who had worked on computation from continuing to work on computation.

yyyk• 6 years ago

1950s-1980s America had complexity, they just overcame it, and you're used to this. Having a notion of 'case' or contractions at all (plenty of languages don't), the imperial system, the odd MM/DD/YYYY format.

nfoz3• 6 years ago

— How about the left angle bracket "<"? Is that at least a "bracket"?
No.
— Why not?
Because it's a LESS-THAN SIGN.

It's petty, and there are bigger problems, but this is one of my main gripes against SGML and its successors.

tingletech• 6 years ago

you don't have to use LESS-THAN and GREATER-THAN in an SGML https://en.wikipedia.org/wiki/Standard_Generalized_Markup_La...

nfoz3• 6 years ago

That's cool! But it's also somehow even worse, by being so complicated.

gumby• 6 years ago

TBF Unicode contains quite a few absurdities because of its commitment to round trip consistency. That means glyphs which were obviously representations of the same thing (such as ‘[‘ in one character set and ‘[‘ in another) got different code points, The names often reflect that too.

nfoz3• 6 years ago

? My problem is not with Unicode here. Unicode has angle-brackets. But SGML (which tbf predates Unicode by a lot) doesn't use them.... It uses less-than/greater-than signs and pretends that they're angle-brackets. They're not. Consequently, HTML <tags> are typographically incorrect and look horrible.

It's made even worse by "programmer fonts" that make the glyphs for less-than/greater-than look like angle-brackets, which is just wrong; now there's no way to write the comparators.

I guess we all just wish that ASCII (and/or keyboards) had one or two more pairs of brackets.

iokevins• 6 years ago

Pronounced Bidi like MIDI, or Bidi like Wifi, or (?)

staktrace• 6 years ago

Like MIDI

kbutler• 6 years ago

Bye-dye (Bi-directional).

Some people/locales pronounce wifi as "wiffy".

zeveb• 6 years ago

Or like bidis, the wonderful flavoured cigarettes banned by Mr. Obama's execrable FDA in 2014, to rhyme with 'needy.'

Man, I miss them.

minikites• 6 years ago

I tried to find what "Left S-shaped bag delimiter" was used for and all I found was the Wikipedia page listing the symbol among other math symbols: https://en.wikipedia.org/wiki/Miscellaneous_Mathematical_Sym.... Does someone here have an explanation?

sp332• 6 years ago

While it does not use this notation, I think this page is the right idea. https://en.wikipedia.org/w/index.php?title=Bag_(mathematics)

jdlshore• 6 years ago

This is surprisingly hilarious for a "Unicode Technical Note." It changed my opinion about the Unicode Consortium—positively!—until I read this:

These technical notes are independent publications, not approved by any of the Unicode Technical Committees, nor are they part of the Unicode Standard or any other Unicode specification. Publication does not imply endorsement by the Unicode Consortium in any way.

Still, awesome and fun. And I learned something.

msoad• 6 years ago

Ok it is 2020. Let me see if I can use brackets in my text:

This is English with some فارسی(Persian) in it. Persian is also called Farsi(فارسی).

این متن به فارسی )Persian) نوشته شده

It worked but I was very confused and I'm not actually sure if position of (Persian) is right. I put it after the Farsi.

The majority Farsi line is completely butchered. Not sure if it is because this text input doesn't support RTL?

jfk13• 6 years ago

That line has two occurrences of U+0029 RIGHT PARENTHESIS in it. I think one of them should have been U+0028.

On top of that, you really want dir=rtl on the element to give it right-to-left base directionality. If that's not an option (e.g. in a HackerNews comment), you can surround it with Unicode directional control characters U+202B RIGHT-TO-LEFT EMBEDDING .. U+202C POP DIRECTIONAL FORMATTING to make it behave correctly as a right-to-left run within a left-to-right context.

LordHeini• 6 years ago

Parsing brackets is even more entertaining in physics due to the bra-ket notation:

https://en.wikipedia.org/wiki/Bra%E2%80%93ket_notation

cormullion• 6 years ago

"U+FDE3" should be U+FD3E, ORNATE LEFT PARENTHESIS.

pabs3• 6 years ago

(2014)

akdor1154• 6 years ago

What I missed in the article, was: why? Why does a text rendering system or encoding conversion system need to care about which brackets are paired?

contravariant• 6 years ago

Of course mathematicians would also consider brackets used like ]0,1[ and [0,1[ as valid notation for open or half-open intervals. And if you try hard enough you can even make your braces {} look like a ξ and write some monstrosity like {ξ∈[0,1[}.

Bidi Brackets for Dummies

Related Stories