You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Ah OK so are these issues where syntax alone would help?
Parsing it is probably significantly easier than running it. Even though that turns a parse error into a runtime error, it could be OK (could hide it between parse_strict or something). I still feel like it is a poorly designed feature but parse support could be reasonable
This is some more evidence that people use it, which is helpful
At least in resholve's case, syntax support alone would help.
I admit I'm a little unsure about this. I wouldn't want it to be a subtle trap for OSH users (especially if it globally softened always-bad brace syntax to a runtime error?)
Since OSH doesn't support coproc at all for now, I guess there might be a positive UX outcome for being able to tolerate the syntax in the context of a coproc: OSH could raise a coherent error that points users to documentation on the state of coproc, regardless of which bash form they're using?
The parser falls over on bash's braced coproc form:
I brought this up previously on zulip (https://oilshell.zulipchat.com/#narrow/stream/121540-oil-discuss/topic/coproc.20syntax.3F) so I'm not really expecting a change of policy here. Opening an issue to document and cross-link, since I finally had a resholve user other than myself encounter/report this :) abathur/resholve#79
Here's the reporter's example: https://github.com/tejing1/nixos-config/blob/45d10ed6c530f80abf19e7c8e1cb028de3ebdb92/homeConfigurations/tejing/email/mailwatch.sh#L15
Seeing their report also reminded me that I had my own case where I wanted to use the named form (but avoided it): https://github.com/abathur/lilgit/blame/main/lilgit.bash#L3-L7
The text was updated successfully, but these errors were encountered: