Permalink
Commits on Sep 4, 2011
  1. ignore vendor bro

    brianmario committed Sep 4, 2011
Commits on Aug 10, 2011
  1. more readme fixes

    brianmario committed Aug 10, 2011
  2. fix comment

    brianmario committed Aug 10, 2011
Commits on Apr 20, 2011
  1. Revert "oh for fucks sake..."

    brianmario committed Apr 20, 2011
    This reverts commit 662e57b.
  2. oh for fucks sake...

    brianmario committed Apr 20, 2011
  3. forgot one

    brianmario committed Apr 20, 2011
  4. convert readme to markdown

    brianmario committed Apr 20, 2011
  5. dammit

    brianmario committed Apr 20, 2011
  6. :(

    brianmario committed Apr 20, 2011
  7. lets try that AGAIN

    brianmario committed Apr 20, 2011
  8. formatting readme

    brianmario committed Apr 20, 2011
Commits on Apr 14, 2011
Commits on Mar 9, 2011
  1. move to 0.2.3

    brianmario committed Mar 9, 2011
  2. after re-thinking the encoding strategy for 1.9, we'll now not even c…

    brianmario committed Mar 9, 2011
    …are about Encoding.default_internal and just return the output string in the encoding the input string was in. This is consistent with how most (all?) of the existing pure-ruby routines work anyway
Commits on Feb 25, 2011
  1. move to 0.2.2

    brianmario committed Feb 25, 2011
Commits on Feb 23, 2011
  1. Rubinius compatibility

    hosiawak committed Feb 23, 2011
Commits on Feb 21, 2011
  1. bump to 0.2.1

    brianmario committed Feb 21, 2011
Commits on Feb 10, 2011
  1. just so we're consistent

    brianmario committed Feb 10, 2011
  2. peace VERSION file

    brianmario committed Feb 10, 2011
Commits on Feb 8, 2011
  1. fix typo in gemspec

    brianmario committed Feb 8, 2011
Commits on Jan 27, 2011
  1. avoid checking leftover length on unescaping

    Eric Wong
    Eric Wong committed Jan 27, 2011
    Ruby always null terminates strings for compatibility with
    standard C string functions, so as long as our IS_HEX() macro is
    working properly and failing on '\0', there is no need to check
    the length of the leftover string.
  2. workaround rb_cvar_set arity change between 1.8 and 1.9

    Eric Wong
    Eric Wong committed Jan 27, 2011
    The extra argument is a warning variable and removed in 1.9
  3. remove unnecessary method calls in HtmlSafety

    Eric Wong
    Eric Wong committed Jan 26, 2011
    The C escaping function already reads the internal value if
    nothing is passed, so the extra Ruby method dispatch is
    wasteful.
  4. avoid rb_funcall for checking the html_secure flag

    Eric Wong
    Eric Wong committed Jan 26, 2011
    This gives a small speedup (larger for small strings) for escaping HTML
    as there's no need for an extra method dispatch.  I'm preserving the
    actual class variable for compatibility, but it could be removed if
    nobody is depending on it.
  5. remove all malloc+free pairs

    Eric Wong
    Eric Wong committed Jan 26, 2011
    For unescapes, this reduces data copies and in the likely case
    where the majority of characters do not need to be unescaped,
    memory efficiently is not greatly impacted.
    
    For escapes, this seems to help a small bit on escaping where
    the output string is longer, too.  rb_str_resize() will release
    memory back to the malloc() pool if the disparity between the
    buffer size and string length is large enough (1024 bytes in
    Ruby 1.9), so memory efficiency for escapes is not overly
    impacted.
    
    The removal of direct malloc() calls (vs xmalloc()) also
    protects us (as best as possible) from out-of-memory errors,
    but will also cause the GC to be invoked more frequently.
  6. reduce memory usage of temporary buffer

    Eric Wong
    Eric Wong committed Jan 26, 2011
    We were allocating 4-8 times the necessary size for the temporary
    buffer.
    
    Now we just have to make sure our maximum size calculations are 100%
    correct as we have no room for error when it comes to buffer
    overflows.