[dmd-internals] Status of dmd pull requests

[Jonathan M Davis]

On Sunday, September 04, 2011 17:19:18 Jonathan M Davis wrote:
> On Sunday, September 04, 2011 17:05:20 Daniel Murphy wrote:
> > Sydney (+10 ?)
> > ~$ cat /etc/timezone
> > Australia/NSW
> 
> *Sigh* It looks like something is messed up with std.datetime dealing with
> that timezone. It looks like I get to go bug hunting...

LOL. Ouch. _That_ was dumb of me. I was trying to get the order of several 
time zones (including local time) in the tests, and somehow I managed to 
forget that diffing times ignores time zones (as it should). So, I subtracted 
the current time in UTC from the current time in local time (which inevitably 
is a matter of microseconds, because it's just the difference between two calls 
to Clock.currTime). The result is inevitably positive. West of UTC (as I and 
the autotester are) needs the result to be positive to order the time zones 
correctly, whereas east of UTC (as you are) needs the result to be negative - 
but it's positive. So, the time zones are in the wrong order for you, and the 
results are wrong. _This_ is why I like it when unit tests are dead stupid 
rather than trying to be clever and concise. It's so easy to screw them up 
when you try and be clever and concise. But no one likes the large number of 
dead stupid unit tests in std.datetime, so I've been trying to make them 
clever and concise. Bleh.

Oh well. At least I know what the problem is now. It shouldn't take very long 
to fix it.

- Jonathan M Davis