Skip to content
Commits on Mar 13, 2013
  1. [maven-release-plugin] prepare release core.logic-0.8.0-rc3

    Hudson @ build.clojure.org committed
  2. @swannodette

    bump version

    swannodette committed
  3. @swannodette

    include 112 test case

    swannodette committed
  4. @swannodette

    optimization

    swannodette committed
  5. @swannodette
  6. @swannodette

    LOGIC-112: Incorrect results with tabled resolution

    swannodette committed
    The bug was introduced when we switched the answer cache to sets. In
    `reuse` we were calling `first` on the set which of course won't
    work.
    
    We not have an `AnswerCache` type. This holds both the list of answers
    for determining the fixpoint as well as the answers as a set for
    quickly determining whether we've already cached an answer.
    
    The additionally memory overhead does make it desirable to have more
    sharing of tabled information between answer caches.
Commits on Mar 12, 2013
  1. @swannodette

    first pass at new table cache type. nondeterminism and general bad

    swannodette committed
    behavior in tabling results from the `(first ansv*)` line in the
    implementation of `reuse` - there's no guarantee what that will return
    for sets. Ooops.
  2. @namin
  3. @namin
  4. @swannodette
  5. @swannodette

    update project.clj

    swannodette committed
Commits on Mar 11, 2013
  1. @namin

    whitespace changes

    namin committed
  2. @namin

    a few little cleanups

    namin committed
  3. @namin
  4. @namin
  5. @namin
  6. @namin
  7. @namin
  8. @namin
  9. @swannodette

    remove stray field access

    swannodette committed
  10. @namin

    Fix copy-term.

    namin committed
    - build-term for vars seemed out-of-date,
    - build-term needs to be implemented for all collections (including
      vectors), not just seq.
Commits on Mar 10, 2013
  1. @swannodette
Commits on Mar 2, 2013
  1. @swannodette

    DRY

    swannodette committed
Commits on Mar 1, 2013
  1. @swannodette
  2. @swannodette

    quick fix, don't penalize update-dom usage in nominal.clj. It's not

    swannodette committed
    clear to me that ::nom information needs to be in the var dom info.
Commits on Feb 28, 2013
  1. @swannodette

    full support for entanglement. merge-with-root now ensures that any

    swannodette committed
    vars entangled by the root and the repointed var will have their
    domains updated to the merged domains of the root. we could probably
    make this a bit more efficient but this is enough to remove any
    reference from nominal.clj of FD or any other constraint domain like it.
  2. @swannodette
  3. @swannodette

    because of var repointing eset may not reflect latest, account for

    swannodette committed
    this when updating entangled vars.
  4. @swannodette

    merge-subst-vals -> merge-with-root. this fn now returns substitutions

    swannodette committed
    in prep for entanglement related changes.
  5. @swannodette

    fix subst-val to accept eset

    swannodette committed
  6. @swannodette
  7. @swannodette
  8. @swannodette
Commits on Feb 27, 2013
  1. @swannodette
  2. @swannodette
Something went wrong with that request. Please try again.