Permalink
Commits on Feb 18, 2010
  1. @schwern

    Version 20100218

    schwern committed Feb 18, 2010
Commits on Feb 16, 2010
  1. @schwern

    Version 20100216.1421_04

    schwern committed Feb 16, 2010
  2. @schwern
  3. @schwern

    whitepace

    schwern committed Feb 16, 2010
  4. @schwern

    10.6's gmtime() loses a day before -70546986201600

    Bug reported to Apple as 7654647.
    schwern committed Feb 16, 2010
  5. @schwern

    Move the choice of asctime format out of the configuration.

    Its not really a config option, but a derivative.
    schwern committed Feb 16, 2010
  6. @schwern

    The test for finding functions (like timegm()) did not really work.

    Unresolved symbols do not always error at compile time, but they
    appear to do so at link time.
    schwern committed Feb 16, 2010
  7. @schwern
Commits on Feb 15, 2010
  1. @schwern

    Version 20100214.1823_03

    schwern committed Feb 15, 2010
  2. @schwern
Commits on Feb 13, 2010
  1. @schwern

    Version 20100213.0504_02

    schwern committed Feb 13, 2010
  2. @schwern
  3. @schwern

    Version 20100213.0000_01

    schwern committed Feb 13, 2010
  4. @schwern

    Update build dependencies

    Test::More for explain()
    
    Need JSON::XS to build
    schwern committed Feb 13, 2010
  5. @schwern

    Give the past localtime() tests an epsilon.

    Can't predict exactly what a system's localtime() will do with a distant date.
    Past dates tend to be calcualted by solar noon, but slightly different system by system.
    schwern committed Feb 13, 2010
  6. @schwern
  7. @schwern

    Typo

    schwern committed Feb 13, 2010
  8. @schwern

    mktime64() now has an accurate idea of what the system mktime can do.

    Let check_max see the time.h configuration before it checks the limits
    
    Split time64_config.h into time64_limits.h so check_max can use time64_config.h
    
    Time::y2038 has to create time64_config.h before it can run check_max
    schwern committed Feb 13, 2010
  9. @schwern
  10. @schwern

    Make the time.h capability check quiessent.

    Also let the user know there may be C errors.
    schwern committed Feb 13, 2010
  11. @schwern
  12. @schwern

    Output check_max as JSON so other projects can use it easier.

    Rework time64_config.h to take a pile of constants representing the mktime/timegm
    limits.  This sucks, but I don't know how to put a struct into a header file without
    C complaining and I don't want to split the config.
    schwern committed Feb 13, 2010
Commits on Feb 10, 2010
  1. @schwern
Commits on Feb 8, 2010
  1. @schwern

    Turn on some more warnings suggested by "An Introduction to GCC" and …

    …fix them.
    
    Also fix that pesky "constant is unsigned in ISO C" warning.
    schwern committed Feb 8, 2010
Commits on Feb 7, 2010
  1. @schwern
  2. @schwern
Commits on Feb 5, 2010
  1. @schwern

    Use a constant for time64 object files.

    For a moment there was more than one.  Nicer to have it in a variable anyway.
    
    Also t/mktime64.t wasn't getting run.
    schwern committed Feb 5, 2010
  2. @schwern
  3. @schwern
  4. @schwern

    Ignore the check_max binary

    schwern committed Feb 5, 2010
  5. @schwern
Commits on Feb 4, 2010
  1. @schwern

    Add a check for timegm and mktime's limits.

    This is necessary for a more accurate mktime.
    schwern committed Feb 4, 2010
  2. @schwern
  3. @schwern

    minor typo in a comment

    schwern committed Feb 4, 2010
Commits on Jan 30, 2010
  1. @schwern

    Add some sanity tests for distant years.

    Perl 5.11 broke y2038 but the tests didn't notice.
    schwern committed Jan 30, 2010