improve error message for using ** #34706
Conversation
based on https://discourse.julialang.org/t/syntax-use-instead-of-and-syntax-is-not-a-unary-operator/34258/5, it seems that while well intentioned, this error message might be causing more confusion than it fixes. I took a stab at addressing it, but I think possibly the better change would be to parse `**` as an operator, and then we could define specific methods that threw better error messages. Specifically, `a**b` should have the error, but if we see `**a`, we should instead be telling people to use splatting, as that is more likely what they were trying to do.
also gives direct advice in the case of splatting.
|
Also, how do I make it so commits don't re-run ci. I know I'm being wasteful here, but I thought that putting [ci skip] in the title would fix it. |
You need to put it in the commit message, not in the PR title. But this is not the kind of PR that should skip CI anyway, your not just updating spelling in the docs, you are changing the parser and if you had made a typo and forgotten a closing quote etc CI would catch it. |
|
Yeah, definitely do not skip CI for this PR/these commits. |
in CI so indeed, not really a good PR for ci skip. |
|
You'd really think I'd be able to edit a single string without typos. Oh well. Fixed. |
|
Thanks for merging this!. |
based on https://discourse.julialang.org/t/syntax-use-instead-of-and-syntax-is-not-a-unary-operator/34258/5, it seems that while well intentioned, this error message might be causing more confusion than it fixes. I took a stab at addressing it, but I think possibly the better change would be to parse
**as an operator, and then we could define specific methods that threw better error messages. Specifically,a**bshould have the error, but if we see**a, we should instead be telling people to use splatting, as that is more likely what they were trying to do.