Closed lishaduck closed 4 months ago
Sorry I had to revert this for now
It causes some quoting issues, and "false"
is actually treated as a string in this context (not a bool) in github actions vars which cause the workflow to misbehave which is why the empty string were returned as they're always falsy in GA.
Also some fields explicitely are sometime explicitely set to false
(like npm
) which cause the comparison to "null" to be incorrect
I tried to troubleshoot it a bit but I gave up for now as I don't have much free time right now
It causes some quoting issues, and
"false"
is actually treated as a string in this context (not a bool) in github actions vars which cause the workflow to misbehave which is why the empty string were returned as they're always falsy in GA.
Huh. GH said it coerces them. I really need to reread those docs (I'm working on a CI PR with quoting issues for another project).
Also some fields explicitely are sometime explicitely set to
false
(likenpm
) which cause the comparison to "null" to be incorrect
Yeah, I should've thought about that.
I tried to troubleshoot it a bit but I gave up for now as I don't have much free time right now
I can take a look sometime. Thanks for giving it a shot.
This is ~1.7× faster on my laptop. CI machines are weak, so it's probably a bigger difference there.
I'm running a lot of vscode and Chrome windows, so hyperfine complains, but it's at least a ~1.2× increase, up to a ~2.6× increase (which was reproducible for enough of a while I'd count this as a win).