Small release of the lambdaisland/deep-diff2 library (recursively diff data, and pretty-print the result).
This fixes the behaviour of ddiff/minimise when inserting/deleting map entries.
https://github.com/lambdaisland/deep-diff2/releases/tag/v2.13.231
Clojure 1.12.5-alpha1 is now available: • https://clojure.atlassian.net/browse/CLJ-2945 - reify - incorrectly transfers reader metadata to runtime object • https://clojure.atlassian.net/browse/CLJ-2228 - constantly - unroll to remove rest args allocation The reify change avoids transferring :line / :col metadata from the reify form to the reified runtime object (which was unintended). reify now behaves like fn in NOT transferring reader metadata (but user metadata on the form is still transferred per the docstring). The benefit of this change is to reduce compiled class size and runtime allocations for reified objects. We have done some poking around to find any unexpected effects from this, but we would appreciate feedback from those working on editors, linters, or other tools that might somehow rely on this.
Clojure 1.12.4 (before)
user=> (meta (reify java.io.Serializable))
{:line 1, :column 7}
user=> (meta ^{:a 1} (reify java.io.Serializable))
{:line 1, :column 7, :a 1}
Clojure 1.12.5-alpha1 (after)
user=> (meta (reify java.io.Serializable))
nil
user=> (meta ^{:a 1} (reify java.io.Serializable))
{:a 1}FYI: we have 1.12.5-alpha1 in production with no issues (I meant to mention that last week but it was a busy week).
if that helps, I don't recall any place in clojure-lsp or clj-kondo that does that for reify :)
No regressions in our test suite at work, so I'll put it up on QA and see if anything pops up (unlikely, I think).
oh hey, if you're merging the change to constantly, may i suggest a cool new patch that does the same for eduction? 😇
probably not the same issue @nbtheduke