Permalink
Commits on Jan 8, 2013
  1. [maven-release-plugin] prepare release core.logic-0.8.0-rc1

    Hudson @ build.clojure.org committed Jan 8, 2013
  2. LOGIC-102: nom/hash tweaks.

    namin committed with swannodette Jan 7, 2013
Commits on Jan 7, 2013
  1. LOGIC-100

    namin committed with swannodette Jan 7, 2013
  2. LOGIC-101

    namin committed with swannodette Jan 7, 2013
Commits on Jan 5, 2013
  1. Add `seqc` constraint

    swannodette committed Jan 5, 2013
Commits on Jan 4, 2013
  1. LOGIC-95: fix some issues with disequality.

    1. -relevant-var? is broken for !=c because it compares vars that have
       been rooted with vars that haven't been, when called from udpatec.
    
    2. Because of nested vars, we need to go deeper when recovering vars,
       and only compare ground terms with not= when invoking a !=c
       constraint.
    
    TODO: for 1, we need to rethink how to maintain the constraint map in
    face of constraints that change the associations between vars and
    constraints.
    namin committed with swannodette Jan 4, 2013
Commits on Jan 3, 2013
  1. remove stale comment

    swannodette committed Jan 3, 2013
  2. LOGIC-53: core.logic converts defrecords

    records must extend `IUninitialized`. They always trump maps.
    swannodette committed Jan 3, 2013
  3. clean up disequality reification, test for reification of disequality

    constraints even if constraints vars don't appear in the answer term.
    swannodette committed Jan 3, 2013
  4. LOGIC-88: disequality reification is broken

    Was not emitting the entire prefix was all.
    swannodette committed Jan 3, 2013
Commits on Jan 2, 2013
  1. fix typo, 77 -> 91

    swannodette committed Jan 2, 2013
  2. LOGIC-91: `predc` constraint entailment issues

    The problem is that cgoal calls `runnable?`, runs the constraint, then
    calls `relevant?`, however when a constraint is already in the store
    we call `relevant?` *first*, because of other constraints may have
    done the work for us.
    
    Added `irelevant?` predicate. `cgoal` and the `run-constraint` now use
    this. Remove all `IRelevant` implementations that simply return
    `true`.
    
    The all different constraint will still exhibit the issue as
    FDConstraint implements `IRelevant` for delegation purporses. We
    should probably eliminate `FDConstraint` and implement some kind of
    templating facility to avoid the boilerplate.
    swannodette committed Jan 2, 2013
Commits on Jan 1, 2013
  1. [maven-release-plugin] prepare for next development iteration

    Hudson @ build.clojure.org committed Jan 1, 2013
  2. [maven-release-plugin] prepare release core.logic-0.8.0-beta5

    Hudson @ build.clojure.org committed Jan 1, 2013
  3. bump version

    swannodette committed Jan 1, 2013
  4. update changes

    swannodette committed Jan 1, 2013
  5. LOGIC-90: <=fd constraint could diverge

    The `-relevant?` implementation for `<=fdc` was wrong. If we have two
    singleton domains we just check that they are not <= to each
    other. However we can also discard the constraint if we have
    `interval-<`. Added test case.
    swannodette committed Jan 1, 2013
Commits on Dec 31, 2012
  1. tweak

    swannodette committed Dec 31, 2012
  2. update readme

    swannodette committed Dec 31, 2012
Commits on Dec 29, 2012
  1. fully lazy core.logic :)

    occur check is now set via field in Substitution, `oc`. `solve` ->
    `-run` and it takes an extra parameter now to set occurs-check. change
    the run macros to use `-run`. remove `lazy-run` no longer needed.
    
    update the benchmarks in bench.clj to use `doall` so we can time
    things properly.
    swannodette committed Dec 29, 2012
  2. bump beta number

    swannodette committed Dec 29, 2012
  3. [maven-release-plugin] prepare for next development iteration

    Hudson @ build.clojure.org committed Dec 29, 2012
  4. [maven-release-plugin] prepare release core.logic-0.8.0-beta4

    Hudson @ build.clojure.org committed Dec 29, 2012
  5. nope was right

    swannodette committed Dec 29, 2012
  6. oops wrong version

    swannodette committed Dec 29, 2012
  7. bump for next version

    swannodette committed Dec 29, 2012
  8. update timing.

    add example from Ciao paper on CLP compilation to JS, looks like our
    disitinctfd constraint compares favorable to Ciao in this case.
    swannodette committed Dec 29, 2012
  9. tweak description

    swannodette committed Dec 29, 2012
  10. update CHANGES.md

    swannodette committed Dec 29, 2012
  11. finally tabled goals that reset between runs!

    `Substitution` type now has a new field `ts`, this holds an atom
    containing a hashmap. Normally this is nil, add `tabled-s` constructor
    function that will create a `Substitution` with this field
    set.
    
    `solve` now uses `tabled-s`.
    
    Removed old `table` fn, no longer needed.
    
    Fix `tabled` macro. Much more disciplined `swap!` usage. `tabled`
    goals macroexpand into fns with generated ids, these are used for the
    lookup in the `ts` field of `Substitution`. Related `swap!` fix to
    `master`.
    swannodette committed Dec 29, 2012