Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Properly calculating time differences in JavaScript (synyx.de)
26 points by VMG on Nov 5, 2012 | hide | past | favorite | 13 comments


This is good info.

I'm a fairly intelligent guy, can follow the plots of Primer, Terminator, and other scifi time travel movies. I have a decent understanding of relativity, spatial reference frames, etc.

But any time I have to deal with slightly more-than-simple time issues in any language, my brain starts more or less falling to pieces. Python and js have good libraries that help, but its still a confounding bugger of an issue.


It's pretty simple once you know the rules:

- Always store times and calculate differences/offsets in UTC.

- Never attempt to convert a local time to UTC (it is not a one-to-one mapping!).

- Depending on your application, convert any dates to a given timezone at the moment they are displayed.

Now Javascript throws a monkey-wrench into the whole affair, because it assumes Dates should be handled in the user's local timezone, which, if you're dealing with timezone issues, is certainly not what you want. So attempting to convert a Date object from one timezone to the next is surely impossible to do without introducing bugs unless you understand the rules for every timezone you're dealing with.

Luckily, a guy named Olson created a database of timezone rules for us mortals to use in our timezone-aware applications. In another stroke of good fortune, some smart people wrote a Javascript library [1] that parses this database for you and can convert your Javascript Date object between any timezone on the planet. So you do all your calculations using this Date object in UTC mode, then convert to the desired timezone at the time they are displayed.

[1] https://github.com/mde/timezone-js


The most i've ever hurt my brain was reasoning about a timezone-aware fleet management system, and what it meant when a vehicle was driven from one timezone to the next while properly billing the time driven. We ended up going with an 80% solution that we know is wrong in some cases, but that we an actually explain to users.


Nitpicking, but this:

Luckily, there is, and that place is UTC. UTC is a time measuring system that does not have daylight savings time and where midnights always are 24 hours apart.

isn't strictly true, at least in the sense that 24 hours isn't always the same length of time because of leap seconds.


The ECMA-262 luckily(?) ignores leap seconds:

http://www.ecma-international.org/ecma-262/5.1/#sec-15.9.1.1

Time is measured in ECMAScript in milliseconds since 01 January, 1970 UTC. In time values leap seconds are ignored. It is assumed that there are exactly 86,400,000 milliseconds per day.


Ignoring leap seconds will not make them disappear :-) They belong to UTC not to ECMA-262.

Therefore, whenever we compute the difference between two UTC times, we should really account for any leap second.


For the purpose of calculating milliseconds between midnights of different days, you can however assume them to be multiples of 862e5 in JavaScript Date.UTC.

You just have to keep in mind that JavaScript UTC != Real UTC

Edit: clarification


No, you cannot. If you do it, your UTC time difference might be one or more seconds off. And this is not what you want, if you need to rely upon that quantity.

Only one Coordinated Universal Time exists: it is the UTC standard. ECMA-262 is the specification of a scripting language, with tons of interesting features but with a very poor understanding of what UTC is.

UTC is based on TAI. Whenever we need to compute time differences of UTC times, we need to account for leap seconds implementing TAI. Implementations are available in a number of languages and very easy to port: - C http://cr.yp.to/libtai.html - Haskell http://hackage.haskell.org/packages/archive/time/1.4/doc/htm... - Erlang https://github.com/secyoure/taider

Full disclosure: author of the Erlang TAI library here.


I completely agree - but as you point out, JavaScript doesn't use proper UTC but a simplified version that does not account for leap seconds. Let's call it Date.UTC.

In Date.UTC, there are no leap seconds and therefore the Day == 862e5 assumption holds and we can use it to calculate the difference between two dates measured in days, which is the point of the blog post.

When doing arithmetic with real UTC, you are absolutely correct and we have to take leap seconds into account.


If you don't mind dependencies, there's Moment.js

http://momentjs.com/


...Which unfortunately falls short when you need to do anything more complicated than transforming dates between local time and UTC. If you need to, say, display a Date in Pacific Standard Time to a user agent in Helsinki, you absolutely need something that understands the timezone database.

I've found that timezone-js [1] does the job.

[1] https://github.com/mde/timezone-js


When I ran into this problem, I ended up switching over to using the "datejs" library, located here "http://code.google.com/p/datejs/downloads/list. Date.js seems to handle Daylight Savings Time just fine; I don't know if it deals with Leap Seconds.


"As a JavaScript veteran you know that you have to use new Date() instead of Date() because the second one returns a string for some reason"

Uhhhhh...




Consider applying for YC's Summer 2026 batch! Applications are open till May 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: