14:21
<jackworks>
https://github.com/caridy/secure-javascript-environment/issues/76
14:21
<jackworks>
magic behavior of `document.all` is poising sandboxes 🤣
20:49
<devsnek>
did people generally agree that normal completions should be a thing
20:54
<Bakkot>
... as opposed to?
20:55
<Bakkot>
mixing abrupt completions with values of non-completion types?
20:55
<devsnek>
yeah
20:55
<Bakkot>
I dunno about people generally but that is definitely the opinion of a majority of the current editor group
20:55
<devsnek>
that normal completion should be a thing you mean
20:55
<Bakkot>
yeah
20:56
<ljharb>
yes
20:56
<devsnek>
also, you had an issue open saying you were going to tidy up completions
20:56
<Bakkot>
yup, still planning to
20:56
<devsnek>
ok
20:56
<devsnek>
i was trying to refactor completions in engine262 and everything exploded
20:57
<ljharb>
completion values will at some point be appropriately socially distanced from regular values
20:57
<devsnek>
apparently at some point in the last two years i added an escape hatch where Completion(language value) returns that language value
22:37
<rkirsling>
so https://github.com/michaelficarra/proposal-first-class-protocols/issues/3 explains why we would avoid using `interface` for a first-class version of what the spec calls a "protocol", but
22:38
<rkirsling>
how did it come to be that it's, e.g., the "iterator protocol" and not the "iterator interface"?
22:39
<rkirsling>
(I'm guessing it's because there was already this "MOP" notion?)