Another (or same, I dunno, different symptom) v1.10.866 potential issue. I just tried ClojureScript v1.10.866 on https://github.com/cljdoc/cljdoc-analyzer’s metagetta module and it is failing on its unknown tagged literal test. All unit tests pass with ClojureScript v1.10.844.
Exception: clojure.lang.ExceptionInfo: /Users/lee/proj/oss/cljdoc/cljdoc-analyzer-master/modules/metagetta/test-sources-special/metagetta_test_special/unknown_tagged_literal.cljc [line 4, col 38] No reader function for tag uknown-tagged-literal.
{:type :reader-exception, :ex-kind :reader-error, :file "/Users/lee/proj/oss/cljdoc/cljdoc-analyzer-master/modules/metagetta/test-sources-special/metagetta_test_special/unknown_tagged_literal.cljc", :line 4, :col 38}
If cljs dev team aren’t thinking, “oh yeah, I know what that probably is” I can follow up with more details and try to isolate.uknown-tagged-literal
this looks like a spelling error?
Good eye, it does, doesn’t it? Maybe it only happened to work on =< v1.10.844? I’ll take a peek at the code, I am not familiar with this change to metagetta.
It is an inconsequential spello, the test is trying to verify that metagetta can still load source for analysis that has any unknown tagged literal(s).
@lee might be a regression caused by https://clojure.atlassian.net/browse/CLJS-3276, similar to https://clojure.atlassian.net/browse/CLJS-3311
try adding any other form before the tagged literal to confirm. (def x 1)
or so, just one top-level form so the new ns logic doesn't trigger
@thheller thanks for sharing your expertise here! Yes, after adding (def x 1)
before the tagged literal the tests no longer fail. So to me the symptom looked different, but you’ve helped to show the cause is the same. Thanks!
yeah, looks like the reader reading the ns form isn't configured fully. same problem probably exists for literals directly in the ns form which I guess nobody did so far?