Permalink
Commits on May 5, 2010
  1. Failing test for integration in parallel_spec vs parallel_test bin sc…

    …ripts.
    Adam Berlin committed May 5, 2010
Commits on May 3, 2010
  1. @grosser
  2. @grosser

    Version bump to 0.3.6

    grosser committed May 3, 2010
  3. @grosser
  4. @grosser

    merge

    grosser committed May 3, 2010
  5. Using ARGV to gather list of files to run instead of --files.

    Adam Berlin and ZABEZ committed May 3, 2010
  6. Quote options on --test-options actually be able to receive options.

    Adam Berlin and ZABEZ committed May 3, 2010
  7. @grosser
  8. Updated README

    Adam Berlin and ZABEZ committed May 3, 2010
  9. adding an option to multiply the amount of processes running

    Adam Berlin and ZABEZ committed May 3, 2010
Commits on May 1, 2010
  1. @grosser
  2. @grosser

    Version bump to 0.3.5

    grosser committed May 1, 2010
  3. @grosser
  4. @grosser
  5. @grosser

    fix spec

    grosser committed May 1, 2010
Commits on Apr 30, 2010
  1. Adds support for --files parameter. Accepts a list of files to run.

    Adam Berlin and ZABEZ committed Apr 30, 2010
Commits on Apr 18, 2010
  1. @grosser

    correct maksims name in readme

    grosser committed Apr 18, 2010
Commits on Apr 4, 2010
  1. @chrisk

    Fix a typo in the README

    chrisk committed Apr 4, 2010
Commits on Mar 27, 2010
  1. @grosser
  2. @grosser

    Version bump to 0.3.4

    grosser committed Mar 27, 2010
  3. @grosser
  4. @grosser

    fix specs

    grosser committed Mar 27, 2010
Commits on Mar 26, 2010
  1. @kscaldef

    Much more even algorithm for distributing tests between processes

    When there is large variation in how long tests take to run, the old
    algorithm could be very uneven in its distribution.  This replaces it
    with a greedy algorithm that assigns each test to the process
    with the lightest workload.  I don’t think it’s provably optimal,
    but it performs much better.  Example:
    
    Before:
    
    size[0] = 427.216147
    size[1] = 531.661853
    size[2] = 471.283475
    size[3] = 512.337818
    size[4] = 519.610019
    size[5] = 528.156902
    size[6] = 513.50903
    size[7] = 646.904167
    group size 0 = 1
    group size 1 = 2
    group size 2 = 3
    group size 3 = 5
    group size 4 = 9
    group size 5 = 15
    group size 6 = 25
    group size 7 = 231
    8 processes for 291 specs, ~ 36 specs per process
    
    After:
    
    size[0] = 518.829876
    size[1] = 518.829363
    size[2] = 518.838745
    size[3] = 518.835515
    size[4] = 518.836727
    size[5] = 518.831238
    size[6] = 518.838641
    size[7] = 518.839306
    group size 0 = 30
    group size 1 = 35
    group size 2 = 37
    group size 3 = 38
    group size 4 = 38
    group size 5 = 37
    group size 6 = 38
    group size 7 = 38
    8 processes for 291 specs, ~ 36 specs per process
    kscaldef committed Mar 26, 2010
  2. @kscaldef
Commits on Mar 23, 2010
  1. @grosser
  2. @grosser

    Version bump to 0.3.3

    grosser committed Mar 23, 2010
  3. @grosser
Commits on Mar 12, 2010
  1. @grosser

    refactor rake tasks to use system instead of exec to fix `marshal dat…

    …a too short` errors
    grosser committed Mar 12, 2010
Commits on Mar 9, 2010
  1. @grosser
  2. @grosser
  3. @grosser

    Version bump to 0.3.2

    grosser committed Mar 9, 2010
  4. @grosser
Commits on Mar 2, 2010
  1. @grosser
  2. @grosser

    Version bump to 0.3.1

    grosser committed Mar 2, 2010
  3. @grosser