hmm on master

i'm getting a bunch of "did not catch tag:" when I'm going through several libs

so for next.jdbc for example, I'm getting that on: ResultSet, DataSource, Connection, ResultSetMetaData, Integer, java.io.Writer, java.sql.ResultSet, PrepaerdStatement

it'd be nice to have a way to extend what tag-from-meta is switching over

yeah, sure, I forgot to delete that println

I'm continuing in the branch ret-types, but it's a work in progress. I'll report back here when it's a good time to check again!

the branch ret-types can be tested.

I have to clean up that branch some more, and then I'll merge to master

👍

but you can test it right now if you want 🙂

oh why not 😄

need to make sure it works with linter-kondo anyway, right? 😉

one issue I struggled with is that ^String can be nilable, but some specs I wrote expected just a ::string. But to not create any false positives, I basically have to treat almost every type a nilable.

that's why (subs nil 1 10) now doesn't create a warning

it sucks, but at least you still get warning for (subs 1 2 3)

oh sure

that's fair.

It appears that within cond->, I'm not seeing unresolved symbol errors. I took a look through the issues on github but didn't spot anything related - is this a known problem, or would you like me to file a bug?

Appears to be both within the condition and the body

@robertfrederickwarner That's known, some macros aren't fully linted, because I just didn't get around to it yet. Will handle it in a future version.

@gerred I think the nil issue has been fixed now in the ret-types branch. Still have to do some cleaning/maybe optimizing

$ echo '(subs nil 1 1)' | clj-kondo --lint -
<stdin>:1:7: error: Expected: string, received: nil.
linting took 10ms, errors: 1, warnings: 0