Permalink
Commits on Dec 7, 2011
  1. Updated CI tags on linux.

    brixen committed Dec 7, 2011
Commits on Dec 6, 2011
  1. More chewing gum, duct tape, and baling wire.

    brixen committed Dec 6, 2011
    I swear I will finish daedalus once 2.0 is released.
  2. Added rb_enc_mbclen.

    brixen committed Dec 6, 2011
  3. fix while loop in rationalize

    lest committed Dec 6, 2011
Commits on Dec 5, 2011
  1. Removed vendor/onig.

    brixen committed Dec 5, 2011
  2. Use vendor/oniguruma.

    brixen committed Dec 5, 2011
  3. Added oniguruma:sync task.

    brixen committed Dec 5, 2011
  4. Merge pull request #1441 from dlt/master

    argent-smith committed Dec 5, 2011
    improving pt-br documentation
Commits on Dec 4, 2011
  1. improving pt-br documentation

    dlt committed Dec 4, 2011
  2. Favor Bignum#to_s performance for small numbers

    jfirebaugh committed Dec 4, 2011
    Before:
       small Bignum#to_s   171888.5 (±8.8%) i/s -     861800 in   5.054854s (cycle=13900)
      medium Bignum#to_s     1310.6 (±7.6%) i/s -       6517 in   5.004385s (cycle=133)
       large Bignum#to_s       13.4 (±7.5%) i/s -         67 in   5.002315s (cycle=1)
        huge Bignum#to_s        0.1 (±0.0%) i/s -          1 in   7.413067s (cycle=1)
    
    After:
       small Bignum#to_s   339909.5 (±7.1%) i/s -    1713328 in   5.065958s (cycle=25196)
      medium Bignum#to_s     2741.1 (±2.7%) i/s -      13770 in   5.027441s (cycle=270)
       large Bignum#to_s        9.6 (±0.0%) i/s -         48 in   5.005566s (cycle=1)
        huge Bignum#to_s        0.1 (±0.0%) i/s -          1 in   7.823482s (cycle=1)
  3. handle block passed in IO#lines

    lest committed Dec 4, 2011
  4. Implement 1.9 Integer#round

    jfirebaugh committed Dec 4, 2011
    Doesn't follow MRI's goofy NUM2INT RangeError logic.
  5. Precalculate the buffer size for Bignum#to_s

    jfirebaugh committed Dec 4, 2011
    Avoids calling mp_toradix_nd many times for large numbers.
    
    Before: 2**123456.to_s # 6m26s
    After:  2**123456.to_s # 30s
    
    Still a lot slower than MRI though.
Commits on Dec 3, 2011
  1. IO#inspect method for 1.9

    lest committed Dec 3, 2011
  2. Revert "Numeric#fdiv coerces self with #to_f"

    lest committed Dec 3, 2011
    This reverts commit 8eec24f.
    
    It cause Complex#fdiv to fail