2023-07-25
[14:49:09.0111] <ljharb>
how come `new Intl.NumberFormat('en-US', { style: 'unit', unit: 'byte', unitDisplay: 'narrow', notation: 'compact' }).format(1234567891)` prints out `1.2BB` and not `1.2GB`?

[14:59:16.0239] <Chris de Almeida>
because you said you wanted bytes

[15:13:05.0094] <Chris de Almeida>
whoa, that is weird

[15:15:32.0560] <Chris de Almeida>
```
100000000 -> 100MB
1000000000 -> 1BB
10000000000 -> 10BB
100000000000 -> 100BB
1000000000000 -> 1 TB
```

[15:15:43.0473] <Chris de Almeida>
 * ```
100000000 -> 100MB
1000000000 -> 1BB
10000000000 -> 10BB
100000000000 -> 100BB
1000000000000 -> 1TB
```

[15:18:06.0174] <sffc>
You want `new Intl.NumberFormat('en-US', { style: 'unit', unit: 'gigabyte' }).format(1.234567891)`

[15:18:29.0556] <Chris de Almeida>
sure, but if you didn't want to have to do your own unit calc for human readable...

[15:19:17.0295] <Chris de Almeida>
it appears every other expression is what one would expect

[15:19:33.0697] <sffc>
There's a proposal for unit calculations/preferences but it's blocked on (1) user preferences and (2) lack of consensus that units math belongs in the standard library

[15:19:56.0679] <sffc>
currently at Stage 1 and last presented in 2020 I believe

[15:20:57.0537] <sffc>
The `1BB` is because you're printing `1B` in compact notation (`1K`, `1M`, `1B`, `1T`), and then you're attaching a unit to the end

[15:21:53.0779] <Chris de Almeida>
 * it appears every other expression is what one would expect.  edit: stops at TB tho

[15:22:45.0066] <Chris de Almeida>
I don't understand why it's not `1G`, what am I missing here

[15:24:18.0978] <sffc>
https://www.youtube.com/watch?v=kJQP7kiw5Fk has 8.2B Views

[15:28:05.0148] <Chris de Almeida>
so you're saying :
K = thousand
M = million
B = billion
T = trillion
and we are just associating M and T with mega and tera erroneously

[15:29:59.0948] <sffc>
Compact decimal notation != SI prefixes

[15:30:14.0246] <Chris de Almeida>
aye

[15:30:39.0937] <Chris de Almeida>
it's the overloading that causes the confusion.  thanks

[15:30:41.0558] <sffc>
See https://unicode.org/cldr/charts/43/by_type/numbers.compact_decimal_formatting.html for the symbols different locales use

[15:31:47.0297] <sffc>
I see where the confusion comes from. I wouldn't be opposed to a proposal that uses SI prefixes _instead of_ compact decimal notation when style is "unit"

[15:32:19.0609] <sffc>
If someone files an issue, we can track that better :)

[15:33:22.0747] <Chris de Almeida>
wouldn't that be a breaking change at this point?

[15:34:12.0987] <sffc>
A normative change, sure, but not necessarily web-incompatible; unclear if any of this behavior is required by the spec. I think not.

[15:34:53.0006] <sffc>
A normative change could be a one-liner "the compact decimal suffix can depend on style"

[15:40:11.0145] <sffc>
Yeah actually in practice we already have special patterns for notation: compact with style: currency. So it shouldn't be hard to allow special patterns to be used in that way for style: unit as well.

[15:48:22.0188] <sffc>
https://github.com/tc39/ecma402/issues/810

[15:50:21.0346] <ljharb>
if the unit is "byte" i wouldn't expect anything but megabyte, gigabyte, terabyte, etc

[16:02:08.0797] <Chris de Almeida>
yeah, now that I understand what's happening, it makes perfect sense, but I agree, the number of people who would actually want this behavior in this case is near zero


2023-07-26
[08:47:15.0783] <ljharb>
an astute question i was asked:
> The M stands for Million and the K stands for KThousand?

[08:50:38.0424] <Chris de Almeida>
indeed.  they have played us for absolute fools

[08:50:41.0065] <Chris de Almeida>
https://cldr.unicode.org/translation/number-currency-formats/number-and-currency-patterns#h.eradhhuxzqqz

[08:52:22.0063] <Chris de Almeida>
lol, I didn't even notice the `KThousand`

[08:56:59.0890] <sffc>
CLDR just reflects common use. It would be interesting to read a study of the origin of these abbreviations in English and in other languages.

[08:58:10.0171] <Chris de Almeida>
ironically, M, B, and T are probably the outliers

[08:58:41.0678] <Chris de Almeida>
kilo prefix in metric system, etc

[08:59:56.0315] <Chris de Almeida>
although then you get into metric ton vs SI megagram...

[09:01:38.0734] <Chris de Almeida>
but the SI prefix for 10^9 IS giga...

[09:02:20.0347] <Chris de Almeida>
clear as mud

[09:02:39.0741] <sffc>
Looks like Quebecois (fr-CA) has it right: https://github.com/unicode-org/cldr/blob/d1ae244c0ed6084b933646707f99925e26a22e97/common/main/fr_CA.xml#L5841

But it seems they're the only one who uses SI prefixes for compact notation

[09:27:57.0680] <ljharb>
for normal quantities these might be fine. but for bits/bytes almost universal use would be mega/giga/tera

[09:44:27.0024] <Chris de Almeida>
right.  I doubt supporting/changing this would be controversial.  (famous last words notwithstanding)