@borkdude i've been looking through clj-kondo for code related to "Mismatched bracket..." -- at least initially it looks to me like at least some of the functionality you mentioned (also?) lives in your clj-kondo-included rewrite-clj fork. does that sound right?

It lives in the inlined directory

yes, i found your tools.reader fork in the inlined directory and see it is a little different from the official 1.2.2 -- the rewrite-clj fork seems to have more of what you were referring to. i am just asking for confirmation that it makes sense to examine the rewrite-clj portion as well -- i've been reading it too.

I don't remember if rewrite-clj also has changes to this. You should be able to see the relevant commit(s) from @marc-omorain

they were not too long ago

👋

There were two changes that I made: - rewrite-clj was not reporting row + col information in many cases where there was an error in source at the tokenizer stage. - I changed the rewrite-clj error message to refer to both the position of both the opening and closing parens.

@marc-omorain ty for the summary!

https://github.com/borkdude/clj-kondo/blob/875a6bcf660fab60e3037b862edcab23dbc6124a/test/clj_kondo/impl/parser_test.clj#L40-L55

This test exposes a lot of the problem, I believe that running those examples through rewrite-clj will expose the issues. Those were my most reduced repro-cases.

ah nice -- will try running that.