She stares at the screen and the little words of green

[venta]

Yesterday I was writing here about someone reading text on a phone. The text appeared in Arabic, I originally wrote, but the numbers appeared in Roman numerals.

No, wait. I, II ? I don't mean Roman numerals. What's the word for ordinary, normal numerals? They're...

Oh yeah, they're Arabic numerals. I remember.

I mentioned this last night to someone who's just got back from a holiday in Beirut[*]. Aha, she said! But people writing in Arabic don't use those numerals. At least, not in the Lebanon. During a very, very long traffic jam she matched the Arabic numbers (by which I mean the numerals as used in England) on the car licence plates with the Arabic-looking squiggles on the other half of each car's licence plate, and deduced that they use a complete different set of symbols to represent numbers.

Apparently it all depends whether you're using Eastern or Western Arabic numerals, and whether you are East or West of Egypt.

Today, I have learned something.

The corollary to this is that our number system is based on the Hindu-Arabic number system. Like theirs, the number is read left-to-right.

So, if you're reading Arabic text, the text flows right-to-left... until you get to a number. Then it briegly switches to left-to-right. Mmmm.... bi-directional text. Just for those who thought that text layout was just too easy and needed a bit more of a challenge.

Interestingly (by which I mean "according to Wikipedia"), for small numbers they are more-or-less read in the same direction as ordinary text: when reading 25 out loud in Arabic, you effectively read "five and twenty". Sadly, 125 is "one hundred and five and twenty", which seems as wilfully perverse as Americans and their middle-endian dates.

[*] She's a travel journalist. She does things like that.

Comments

[fractalgeek.livejournal.com]

I used to do lots of (minor) localisation, and I tried to organise dialogs in ways that did NOT use sentences for exactly this eg: "File(s) retrieved: %1n" (where %1 is of course a postitional parameter). Anything I could fire through system text translation was good.

Arabic going "1, 2, many" is new to me

My #1 rule: Use ISO 8601 dates, not local for any date passed around as processable text. One system we did, roughly 1/3 of all bugs were local-format dates being processed incorrectly due to system and user locales being different.

Japanese? count in groups of 4 (not 3) - Easy, except that's for Japanese characters, and then do you put in the size counters? Usually in financial contexts, culture prefers western (ie Arabic, etc etc). Multiple overlapping calendar forms whose end date for a reign comes into force retrospectively.

The all-time killer though is what do display for a time where you have to take daylight savings into account.... A repeat appointment, for instance should be booked to the same slot after DST may have been applied. AAARRRGHHHH.

[celestialweasel.livejournal.com]

In China (and Japan?) numbers are grouped in 2 not 3 digits. In India the first grouping is 3 thereafter 2 e.g.

1
10
100
1,000
10,000
1,00,000
10,00,000
1,00,00,000

[thefon.livejournal.com]

Chinese numbers are complicated
http://en.wikipedia.org/wiki/Chinese_numerals

I've seen them grouped in 4 digits (but not 2), because they have single words for 10,000 (万) and 100,000,000 (亿). So the population of China is only 14 of those.

They have yet a third set of numerals, that are even more difficult to write. These are used in financial documents, to make them hard to forge.

So these days, China mostly uses arabic numbers like the rest of the world.

[venta.livejournal.com]

daylight savings

The worst example of failure-to-handle this I've seen in recent times is Skype. I happened to be chatting to someone online (IM, not VoiP) when the clocks went back last autumn.

Skype re-set its internal clock, and for the next hour the conversation got interleaved with the previous hour's conversation. Deeply horrible.

[d-floorlandmine.livejournal.com]

[wince]