Tests broken in Windows / Cygwin setup

[lgrammel]

I am trying to build d3 in a Windows / Cygwin setup. I was able to run the test cases. However, a few of them (mostly timer, epsilon, and date related) are broken (see below). I'll see if I can fix them - any hints on how to adjust them so they work on both Windows & Mac are very welcome :-)

timer with no delay
  ? calls every 17 ms
    » expected 92 to be in within 20 of 51 // timer-test.js:22

timer with a specified delay
  ? calls every 17 ms
    » expected 90 to be in within 20 of 51 // timer-test.js:34

linear invert
  ? coerces range value to numbers
    » expected 0.5001141552511416 to be in within 0.000001 of 0.5 // linear-test.js:155

log invert
  ? coerces range value to number on invert
    » expected 3.1631089809962054 to be in within 0.000001 of 3.1622777 // log-test.js:150

pow range
  ? coerces range values to number on invert
    » expected 0.5001141552511416 to be in within 0.000001 of 0.5 // pow-test.js:62

sqrt range
  ? coerces range value to number on invert
    » expected 0.25011416828256294 to be in within 0.000001 of 0.25 // sqrt-test.js:62

day floor
  ? observes start of daylight savings time
    » expected Sat, 12 Mar 2011 00:00:00 GMT,
    got      Sun, 13 Mar 2011 00:00:00 GMT (deepEqual) // day-test.js:26

day ceil
  ? observes start of daylight savings time
    » expected Sun, 13 Mar 2011 00:00:00 GMT,
    got      Mon, 14 Mar 2011 00:00:00 GMT (deepEqual) // day-test.js:48

  ? observes end of daylight savings time
    » expected Sun, 06 Nov 2011 00:00:00 GMT,
    got      Mon, 07 Nov 2011 00:00:00 GMT (deepEqual) // day-test.js:54

format
  ? formats time zone
    » expected '-0800',
    got      '+0000' (==) // format-test.js:141

hour floor NPT
  ? observes 15-minute offset
    » expected Fri, 31 Dec 2010 17:15:00 GMT,
    got      Fri, 31 Dec 2010 23:00:00 GMT (deepEqual) // hour-test.js:43

hour floor IST
  ? observes 30-minute offset
    » expected Fri, 31 Dec 2010 17:30:00 GMT,
    got      Fri, 31 Dec 2010 23:00:00 GMT (deepEqual) // hour-test.js:50

hour ceil NPT
  ? observes 15-minute offset
    » expected Fri, 31 Dec 2010 18:15:00 GMT,
    got      Sat, 01 Jan 2011 00:00:00 GMT (deepEqual) // hour-test.js:83

hour ceil IST
  ? observes 30-minute offset
    » expected Fri, 31 Dec 2010 18:30:00 GMT,
    got      Sat, 01 Jan 2011 00:00:00 GMT (deepEqual) // hour-test.js:90

hours
  ? observes start of daylight savings time
    » expected [ Sun, 13 Mar 2011 09:00:00 GMT, Sun, 13 Mar 2011 10:00:00 GMT, Sun, 13 Mar 2011 11:00:00 GMT ],
    got      [ Sun, 13 Mar 2011 01:00:00 GMT, Sun, 13 Mar 2011 02:00:00 GMT, Sun, 13 Mar 2011 03:00:00 GMT, Sun, 13 Mar 2011 04:00:00 GMT ] (deepEqual) // hours-test.js:35

  ? observes end of daylight savings time
    » expected [ Sun, 06 Nov 2011 07:00:00 GMT, Sun, 06 Nov 2011 08:00:00 GMT, Sun, 06 Nov 2011 09:00:00 GMT ],
    got      [ Sun, 06 Nov 2011 00:00:00 GMT, Sun, 06 Nov 2011 01:00:00 GMT] (deepEqual) // hours-test.js:42

hours NPT
  ? observes 15-minute offset
    » expected [ Sun, 06 Nov 2011 18:15:00 GMT, Sun, 06 Nov 2011 19:15:00 GMT, Sun, 06 Nov 2011 20:15:00 GMT ],
    got      [ Mon, 07 Nov 2011 00:00:00 GMT, Mon, 07 Nov 2011 01:00:00 GMT, Mon, 07 Nov 2011 02:00:00 GMT ] (deepEqual) // hours-test.js:50

hours IST
  ? observes 30-minute offset
    » expected [ Sun, 06 Nov 2011 18:30:00 GMT, Sun, 06 Nov 2011 19:30:00 GMT, Sun, 06 Nov 2011 20:30:00 GMT ],
    got      [ Mon, 07 Nov 2011 00:00:00 GMT, Mon, 07 Nov 2011 01:00:00 GMT, Mon, 07 Nov 2011 02:00:00 GMT ] (deepEqual) // hours-test.js:59

? Broken » 1597 honored 18 broken (6.480s)

[mbostock]

Sorry, I don't know about Windows. Looks like a combination of floating point differences (which is surprising, since floats are standardized in JavaScript) and timezone problems. The timezone problems are discussed in #718 and #730.