@borkdude Just tried the (macOS) binary you posted above. I hope you forgive me for not going through all that has already been reported and I’ll just report my finding here:

(ns metrics-server.gitlab-fetcher
  (:require [clj-http.client :as client]
            [clojure.string :as str]
            [clojure.data.json :as json])
  (:import java.time.ZoneId
           java.time.format.DateTimeFormatter
           java.time.temporal.TemporalAdjusters
           java.time.LocalDate
           java.time.DayOfWeek))

(def zone-id (. ZoneId systemDefault))
(def formatter (. DateTimeFormatter ISO_OFFSET_DATE_TIME))
(defn date-to-string [date-time]
  (-> date-time
      (. atStartOfDay)
      (. atZone zone-id)
      (. format formatter)))
(def last-monday
  (-> (. LocalDate now)
      (. with (. TemporalAdjusters previousOrSame DayOfWeek/MONDAY))))
(def before-last-monday
  (-> last-monday
      (. with (. TemporalAdjusters previous DayOfWeek/MONDAY))))
(def before-before-last-monday
  (-> before-last-monday
      (. with (. TemporalAdjusters previous DayOfWeek/MONDAY))))

src/metrics_server/gitlab_fetcher.clj:11:17: error: unresolved symbol ZoneId
src/metrics_server/gitlab_fetcher.clj:11:24: error: unresolved symbol systemDefault
src/metrics_server/gitlab_fetcher.clj:12:19: error: unresolved symbol DateTimeFormatter
src/metrics_server/gitlab_fetcher.clj:12:37: error: unresolved symbol ISO_OFFSET_DATE_TIME
src/metrics_server/gitlab_fetcher.clj:15:10: error: unresolved symbol atStartOfDay
src/metrics_server/gitlab_fetcher.clj:16:10: error: unresolved symbol atZone
src/metrics_server/gitlab_fetcher.clj:19:10: error: unresolved symbol LocalDate
src/metrics_server/gitlab_fetcher.clj:19:20: error: unresolved symbol now
src/metrics_server/gitlab_fetcher.clj:20:10: error: unresolved symbol with
src/metrics_server/gitlab_fetcher.clj:20:18: error: unresolved symbol TemporalAdjusters
src/metrics_server/gitlab_fetcher.clj:20:36: error: unresolved symbol previousOrSame
src/metrics_server/gitlab_fetcher.clj:23:36: error: unresolved symbol previous

Also, this code:

(defn extract-relevant-fields [{:keys [key name]
                                [{:keys [value]}] :measures}]
  {:key key
   :name name
   :value value})

src/metrics_server/complexity.clj:7:11: error: unresolved symbol value

Similarly:

(defn extract-relevant-fields [{{:keys [additions deletions]} :stats}]
  {:additions additions
   :deletions deletions})

src/metrics_server/churn.clj:4:15: error: unresolved symbol additions
src/metrics_server/churn.clj:5:15: error: unresolved symbol deletions

@stefan.van.den.oord Thanks for that. Some of those have already been solved, but there are some unknown ones, so I’ll go through them and let you know.

@stefan.van.den.oord I believe most, if not all, errors are now resolved in this binary: https://3614-176829714-gh.circle-artifacts.com/0/release/clj-kondo-2019.06.08-alpha-SNAPSHOT-macos-amd64.zip Feel free to test and report other issues at: https://github.com/borkdude/clj-kondo/issues/174

In the CLI tool, I’m getting an “Unparsable namespace form” for:

(ns foo
  (:require [clojure.string :as str]
            #?(:cljs [cljs.reader :as reader])))

in a .cljc file

If I remove the reader conditional, it works

I can reproduce this if you lint the code as non-cljc, but as clj. but if you put this code in a .cljc file, it works

ok, so it is probably my emacs config that’s broken

to make sure, please test from a shell: clj-kondo --lint /tmp/foo.cljc

clj-kondo could probably report a better error message like: reader conditional not supported in .clj file or something

https://github.com/borkdude/clj-kondo/issues/256

ok, I had the old custom define-flycheck-checker instead of using flycheck-clj-kondo package

if you’re using flycheck-clj-kondo, it should trigger this inside a .cljc file: https://github.com/borkdude/flycheck-clj-kondo/blob/master/flycheck-clj-kondo.el#L74

ok, makes sense

is there (or should there be) a way to mark namespaces that are required for side effects in the source (not config)?

because I think that would be good documentation to be present in the source where the require is placed and could be used for linting purposes

@tatut that’s what we’re trying to achieve in this discussion: https://github.com/borkdude/clj-kondo/issues/241 a convention that all tooling/linters could use

the voting has already started, so please read the issue and bring out your vote 🙂

tl;dr: https://github.com/borkdude/clj-kondo/issues/241#issuecomment-500805941

voted

@tatut if there are many similar namespaces you want to capture in one go, you can also use a regex in the config

@borkdude Confirmed, well done! 🎉

@tatut another idea: https://github.com/borkdude/clj-kondo/issues/241#issuecomment-501693768

that would be enough from a tooling stand point imo, but I kind of like forcing myself to explicitly state it (for documentation purposes)

now the documentation is left to the comments (if any)

but maybe less is more in this case, no need to get everyone to agree on a proper metadata style… just omit :as and :refer and it is assumed to be for side effects

@tatut I usually use comments to document why it is required, it can be for various reasons and comments do give you the freedom to explain it how you want

I think I would still use comments even if there were a metadata standard

should clj-kondo just go ahead and implement it as per your comment? with optional configuration to set it off (something like :assume-implicit-require flag)

I’m all for it, now that I think about it

yeah, I don’t see why not, in the next version

I think it makes sense as the default behavior (underreporting is better than overreporting) and an option to turn it off

@borkdude regarding that #241 issue, probably a “stupid” question, but does it have to be namespaced? Can’t it be simply ^:implicitly_used without ns (for example)?

that has been discussed in the issue, some people think it should be namespaced

just read all the comments, hehe 😉

Right, I did but too quickly apparently, sorry 😉

why the underscore though?

we could also just go with ^:used maybe

Ah my bad, that’s me being too new to clojure

cool I like that a lot, ^:used. Very succinct and to the point.

yeah

having trouble keeping the underscores out of my code as well 🙂 Maybe a new linter rule? 🤪

(prefer-kebab-to-snakes or something)

neh, sometimes you have to use those keys, e.g. when writing to a database table

(even though I’m veggi so I don’t feel comfortable with either)

there is vegan kebab

good one in Berlin: https://www.angloitalianfollowus.com/who-knew-vegan-kebab-meat-exists-at-voner-berlin

I do live close to the German border, but that’s still a bit too far away I guess.

feel free to propose ^:used in the issue

Anyway, what was that about the database table, didn’t quite understand that?

because table column names don’t support hyphens basically

Ah right thanks.

Documentation for an upcoming new unresolved symbols linter: https://github.com/borkdude/clj-kondo/blob/unresolved-symbols/doc/config.md#exclude-unresolved-symbols-from-being-reported If you like to take it for a test run, let me know.