| 07:49 | <Jesse (🇪🇸)> | numerics fans! Shall we do a minisummit while we're here? One idea would be to meet tomorrow after plenary |
| 07:49 | <Jesse (🇪🇸)> | AFAIK a number of us are also here next week for the Web Engines Hackfest |
| 08:13 | <nicolo-ribaudo> | eemeli Time to get the fractionalDigits vs fractionDigits bikeshedding on the agenda! |
| 08:14 | <sffc> | There isn't much extra time this week, but I want us to reserve a nice chunk of time for this. I would be open to a nice Saturday walk to Fisterra with JS Numerics on the agenda. |
| 08:15 | <eemeli> | The community event is tomorrow an hour after the meeting ends, and AFAIK ryzokuken and I are due to talk about MF2 there. |
| 08:16 | <ryzokuken> | yep the community event starts 18:30 tomorrow |
| 08:16 | <Jesse (🇪🇸)> | Friday could also work? at the moment Friday seems to be kind of empty |
| 08:17 | <sffc> | I already have Friday booked for time with the Igalia Intl team about Intl Era Month Code, etc. |
| 08:18 | <eemeli> | Could do a numerics dinner tomorrow after the community event? |
| 08:35 | <Jesse (🇪🇸)> | that might be getting a bit late but I'm open to it |
| 08:41 | <sffc> | I'd prefer during the day on Saturday than after a long day of meetings and community event |
| 08:42 | <sffc> | It seems the TC39 schedule is light on Friday so maybe we could reach out to the chairs to reserve a block for a breakout session on this |
| 08:44 | <Jesse (🇪🇸)> | I'll float the idea to the chairs; there may be some interest in breakout sessions as we did in Seattle |
| 08:48 | <eemeli> | I'd be fine with anything on Friday. Not really up for scheduling work stuff for Saturday; I'm rather looking forward to having some time off. |
| 08:55 | <sffc> | Are y'all going on the city tour today? (I went on it last time I was here) |
| 09:01 | <Jesse (🇪🇸)> | I leaning towards not doing it (I've also already done a tour) |
| 09:01 | <Jesse (🇪🇸)> | (but I'll be at the dinner) |
| 09:28 | <ryzokuken> | I will |
| 09:29 | <ryzokuken> | both options sound good to me and slightly preferable than doing it after the community event |
| 09:33 | <eemeli> | Getting a timeslot within the schedule on Friday sounds like the best outcome. |
| 09:35 | <Jesse (🇪🇸)> | PR set up: https://github.com/tc39/agendas/pull/1891 |
| 09:40 | <eemeli> | I added a slides link PR for keep-trailing-zeros, with content copied almost verbatim from its readme: https://docs.google.com/presentation/d/1gunNRRXJNdDwqTHh-XjV3ueI8PFasRI9WcF4KfWvxE0/edit?usp=sharing |
| 09:43 | <nicolo-ribaudo> | Maybe add an example showing that the proposed behaviour is the same as when you force the .0 through minFractionalDigits |
| 09:50 | <nicolo-ribaudo> | Do you know if there is any language where PluralRules differs between 1.1 and 1.10? |
| 09:51 | <eemeli> | Done, I think -- added slide 6. |
| 09:56 | <eemeli> | Apparently these:
|
| 09:57 | <Jesse (🇪🇸)> | but these differences would be respected in the language-specific plural rules, right? |
| 09:58 | <eemeli> | Yes. |
| 09:58 | <Jesse (🇪🇸)> | iow it would already be buggy, today, and buggy perhaps right after launching this change, but fixable later |
| 10:02 | <eemeli> | My npm package |
| 14:43 | <Jesse (🇪🇸)> | ok we presented our case again |
| 14:43 | <Jesse (🇪🇸)> | congrats on getting preserve-trailling-zeroes to stage 1 |
| 14:43 | <Jesse (🇪🇸)> | ! |
| 14:44 | <Jesse (🇪🇸)> | it seems that there's some non-trivial interest in having a polymorphic amount |
| 14:45 | <Jesse (🇪🇸)> | one thing that I'd like to see are arguments for that, different from the (reasonable, understandable, valid) desire to have a notion of amount that can handle all "numerics" |
| 14:46 | <Jesse (🇪🇸)> | I thought that we walked back from the idea of supporting bigints (at least, bigints with more than 34 significant digits) |
| 14:47 | <Jesse (🇪🇸)> | IIRC the main motivation for that was cryptocurrency |
| 14:47 | <sffc> | That's not the conclusion I drew |
| 14:48 | <ljharb> | the motivation for me is "if it's not tied to one single numeric type, it should handle them all" (to be clear, decimal-only is fine, number-only is fine, but if it's decimal and number i'm going to insist it also include bigint) |
| 14:49 | <Jesse (🇪🇸)> | I'd be interested to know how much slower a polymorphic amount would be compared to a decimal-backed amount |
| 14:49 | <Jesse (🇪🇸)> | I mean, surely it's not faster, and I can see, without digging into any details, why it would be slower |
| 14:50 | <Jesse (🇪🇸)> | we also generally don't position decimal as something fast |
| 14:52 | <Jesse (🇪🇸)> | (I always hesitate to emphasize this becuase I don't want to be misunderstood as being totally insensitive to performance) |
| 14:53 | <eemeli> | Is performance being proposed as a rationalisation for Amount? I thought that it explicitly was not for Decimal as a whole. |
| 14:53 | <sffc> | Polymorphic Amount is probably not that much slower, but my point is that it's not free, and I haven't heard motivation for why we would otherwise prefer Polymorphic Amount over Decimal Amount |
| 14:54 | <Jesse (🇪🇸)> | right, performance isn't being proposed in connection with amount |
| 14:54 | <eemeli> | An Amount supporting all numeric types would not need to be fully polymorphic, like the polyfill it could hold one BigInt and one Number to represent all numeric values. |
| 14:55 | <eemeli> | (bigint for the value, number for the count of fractional digits) |
| 14:55 | <Jesse (🇪🇸)> | ah, a kind of bigdecimal amount? |
| 14:55 | <eemeli> | And then it could have separate toNumber, toDecimal, and toBigInt methods. Which could throw. |
| 14:55 | <eemeli> | Yeah. |
| 14:57 | <eemeli> | In particular considering non-intl use cases, I think supporting the numeric values that we already have in JS with Amount kinda counts. |
| 14:58 | <sffc> | I was in the middle of drafting a message saying that Decimal Amount avoids problems exactly like this |
| 14:58 | <eemeli> | It doesn't really "avoid" the problem, just moves it to happen in the constructor. |