Timezone localization in unit tests is tied to PST about moment

[timrwood]

Yeah, I think the only way to test the issue is to replicate the code for getting the timezone. And that doesn't seem like a good test at all.

Maybe I could store a hash of timezones and match the getTimezoneOffset to them. Something like the following link?

http://blog.redfin.com/devblog/2007/08/getting_the_time_zone_from_a_web_browser.html

[timrwood]

Hmm, after looking into the timezone information here http://www.iana.org/time-zones, it looks like this may be a sub-optimal solution, as the number of lines in the files for 2011 timezone information is over 15000 lines.

The North America file is over 3k lines itself.

Although, as it is for unit tests, the hash of timezones may be the best solution, as it's not that critical to the codebase.

[jgwhite]

Maybe a naive regex would be enough for this unit test? Something like:

var b = moment(new Date(2010, 1, 14, 15, 25, 50, 125));
ok(b.format('zz').match(/^[A-Z]{3,4}$/), 'zz becomes timezone code');

Adjusted appropriately for different locales, this might be enough to verify that something in the ballpark of a timezone code is being returned.

What do you think?

[jgwhite]

I’ve submitted a pull request on this one with my adaptations to the tests.