Error formatting macro: pagetree: java.lang.NullPointerException

Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.


Project/LibraryWhat needs updating?Status
Tests in Clojure itself

Some tests were disabled when the hash changes were first made, with this commit.

DONE (except one test still disabled): Most of them were updated to be independent of Clojure's hash function with this commit, but the following line at or near line 311 in file test/clojure/test_clojure/java_interop.clj is still commented out:

;;(test-to-passed-array-for hash-set)

core.cacheSome tests failed due to hash-dependent sequence ordersDONE: Patch for ticket CCACHE-33 committed
core.matchSome tests failed due to hash-dependent sequence ordersDONE: Fix for ticket MATCH-94 committed
core.matrixMike Anderson says it should be ready for Clojure 1.6 with no updates needed.DONE: No changes needed.
core.rrb-vectorHash calculation needs updatingTicket CRRBV-3 created, fixes in progress.
data.avlHash calculation needs updatingTicket DONE: Fix for ticket DAVL-3 created with proposed patch committed in commit 1, 2, and 3.
data.finger-treeHash calculation for finger trees needs updating for 1.6Ticket DFINGER-2 created with several proposed patches
immutable-bitsetHash calculation for bitsets needs updating for 1.6DONE: Fixed with this commit.  Note that this data structure is a special case, in that it can only contain Long values, thus in Clojure 1.5.1 and earlier it was guaranteed that (.hashCode immutable-bitset) and (hash immutable-bitset) were equal for all immutable-bitsets.  This is not true in general for other collections, because (.hashCode x) and (hash x) are in general different for Integers, Shorts, Bytes, etc.
instaparseMark Engelberg says this relies on incremental hashing, and will need some rethinking on how to update.Some changes have been made on a branch and tested for correctness and performance.  Progress reports here, here.
orderedNamespaces and flatland.ordered.set appear to need updates 
potemkinA map implementation needs updatingZach Tellman plans to fix this soon
tools.namespaceSome tests failed due to hash-dependent sequence ordersDONE: Ticket TNS-16 created, fix comitted.
tools.nreplSome tests failed due to hash-dependent sequence ordersDONE: Fixed with this commit.
usefulflatland.useful.deftype/defmap probably needs updating.  The namespace appears not to need any updates, as it uses Clojure maps but does not customize them at a low level that overrides the hash calculation.