| 09:54 | <HE Shi-Jun> | A question about import defer. Currently, a module include top level-await will not be deferred. Could we allow implementations do some static analysis and optimization to defer the sync part (code after the last await expression) ? |
| 10:45 | <Ashley Claymore> | Wouldn't that break run until completion semantics? |
| 10:45 | <Ashley Claymore> | Maybe not if it, as you say, doesn't start to run |
| 10:46 | <Ashley Claymore> | I'm not sure how big a saving that would add, just the last part of the N preloaded async modules |
| 10:49 | <Ashley Claymore> | Interesting idea though, I'd love to try it out on our version of import defer at Bloomberg but our equivalent of TLA isn't as statically analysable so not really possible |
| 15:00 | <rbuckton> | The tricky part of "defer the sync part" is that TLA can be conditional, with the final await sitting behind a branch that might not be evaluated. You cannot know whether it will be until the condition is evaluated. Which part, then, is the sync part? |
| 16:50 | <saminahusain> | Great meeting and dinner! |