TC39 Module Harmony
2024-08-22
plaintext logs
prev next
source on github
08:21
<jakobjingleheimer>
I'm swamped this week and can't make it. I'll send Fireflies to record and take notes; someone will have to let it into the meet.
08:24
<jakobjingleheimer>
If you want to discuss anything before Fireflies starts, it will wait for 5 minutes after asking to be let in. Message me if you need it triggered to re-join. It takes about a minute to arrive after i trigger it (so it may not be there exactly at 17:00 CEST—if it doesn't show up by ~17:02, message me and I'll re-trigger it)
14:58
<nicolo-ribaudo>
Thank you! I'm in the meeting now, so probably I can let it in whenever you want :)
14:59
<Chengzhong Wu>
I have a conflict first half hour. Will join late, if you are still around
15:00
<jakobjingleheimer>
Thank you! I'm in the meeting now, so probably I can let it in whenever you want :)
Done! LMK if it doesn't join within the next couple minutes
15:03
<nicolo-ribaudo>
Done! LMK if it doesn't join within the next couple minutes
It's in :) I assume it's working
16:22
<jakobjingleheimer>
It's in :) I assume it's working
https://app.fireflies.ai/view/TC39-Module-Harmony::qYPlsVWSkn3qNMqC
19:14
<guybedford>
I've posted a new variant of the reexports analysis in https://github.com/tc39/proposal-esm-phase-imports/issues/20#issuecomment-2305458992, it would be great to get some feedback further to work towards specifying the full analysis.
19:14
<guybedford>
/cc kriskowal
19:21
<kriskowal>

TIL minifier invariance is desired, and that makes sense. I read that as:

assert.deepEqual(
 new ModuleSource(       source ).exports(),
 new ModuleSource(minify(source)).exports());
19:25
<kriskowal>
Starting with the bikeshed, consider naming the determinants/types: export/Export, reexport/Reexport, reexports/Reexports.
19:27
<kriskowal>
I like that this cleanly separates the motivating use cases for imports() and exports() in general, where bindings() conflates those concerns.
19:28
<guybedford>
Starting with the bikeshed, consider naming the determinants/types: export/Export, reexport/Reexport, reexports/Reexports.
Maybe local | external | reexports?
19:28
<kriskowal>
That is imports() is sufficient for traversing the dependency graph and exports() is sufficient for collecting the names and their linkage up the graph.
19:29
<kriskowal>
I dislike local because its dual is remote and absent in the metaphor.
19:29
<guybedford>
Hmm, or something like internal | external | reexports?
19:30
<kriskowal>
The duality of internal and external is satisfying but the incongruence of reexports is dissatisfying.
19:30
<kriskowal>
Distinguish reexport from reexports on the plural is dissatisfying too, but reexport and reexportAll is disatisfying in a different way.
19:31
<kriskowal>
But I would be fine with either reexports or reexportAll.
19:31
<guybedford>
export-all?
19:31
<kriskowal>
And “by fine with”, this is emphatically about pleasure and not sufficiency.
19:32
<kriskowal>
Is there a precedent for kebab-case descriminators? I have a feeling they will end up in object keys, which will have some dissatisfying aethetics imposed by prettier.
19:32
<guybedford>
I mean, it's a Rust-ism, but becoming fairly well-established
19:33
<kriskowal>
I am very pro-kebab for URLS, file-names, and module-specifiers.
19:33
<kriskowal>
Also lunch.
19:34
<kriskowal>
I assume Intl and Temporal are rich in descriminator precedents.
19:34
<kriskowal>
But if we can side-step that particular bike-shed with a more clever treble, that would be a win.
19:51
<kriskowal>
(And I could buy either (export, reexport, reexport-all) or (export, reexport, reexportAll) without qualm.)