Stop asking users for their timezone – use per-device settings and JavaScript
trevoro.net
trevoro.net
Also, it should be noted which time zone a time is in. A naked date/time is about as useful as saying that you're 3 units away from me. What unit does "unit" represent? shrug
2) Usually your times are stored server-side in UTC, so you never have to worry about timezones anywhere. It simplifies things a great deal. "Naked" = UTC. You assume that, whenever it's stored, it's already been translated from the user's timezone when the date was initially generated, or else the time it happened on the server was the actual time.
So far, no complaints.
What brought on this particular use case was a JSON response where all the times were in UTC. I'm not really into forcing a technique on someone when it isn't required. If that's not your use case, then you don't have the problem!
Progressive enhancement is far better for the end user and not all that hard to implement. Some people choose to disable JavaScript, and there are compelling reasons to do this especially if, for example, you're using TBB and want to minimize your attack surface, as has been demonstrated recently.
JavaScript isn't evil by any means, and it's really important for creating cool shit, but it should never be required. If I can't use your site with scripts disabled, neither can a lot of other people. Sucks to be them though, right?
What a ridiculous statement. Lots of things are "required" if you want certain products to be useful. An XBox is required if I want to play XBox games.
Hell you require a web browser just to be able to enjoy the pleasure of checking that "disable javascript" box (though be careful you don't use Mozilla) in the first place (I mean you could use CURL but then you don't actually get to disable javascript).
The point is lets stop kidding around about Javascript here, no-javascript guy.
> If I can't use your site with scripts disabled, neither can a lot of other people.
People who need to disable Javascript are a vanishingly small portion of any potential market, there are many more use cases as far as web apps go where javascript is essential to any decent user experience. Not everything built on web technology is only about content these days. If you don't want to use that technology be my guest, but don't pretend your an important market to people developing on that technology.
> Screen readers and bots (including search engine spiders) have an easier time with well written HTML.
True but two things.
1) Web applications rarely rely on SEO outside of landing pages and a content strategy neither of which should rely on javascript for obvious reasons.
2) Designing applications for disabilities is extremely hard. Much harder than simply using "well written" HTML. Do you know how hard it is for blind people to play most video games? Again we're talking about interactive applications, not simply content. With content there is no problems but as we already went over the web is not only content now.
Edit: One more thought, technically translating the time to the local timezone setting of the browser is "progressively" enhanced. You could always just display UTC time if JS is disabled assuming that was important enough to your users to disable JS (or for bots).
People who need to disable Javascript are a vanishingly small portion of any potential market.
I frequently disable it while browsing casually and, for about a month or so, it was policy at work until we sorted out a few in-house security issues (namely browsing etiquette for some of the folks). Our CMS was broken during this time and the front UI was quickly re-written in plain HTML. After that, we sorta left it that way.
1) Web applications...
Actually, I would. I would be completely surprised if it is common (by that I mean at least more than IE6 usage) at all. However, you've presented no evidence of this fact only anecdote. Seriously there is absolutely no evidence that there is a growing popularity of no-js people out there. Perhaps for very specific demographics, in which case I certainly hope whoever is building product for them knows their customer well enough to know that or God help them.
Either way they are not going to be building the latest in interactive experiences or web-based gaming for your cousin with her rubbish computer are they. That doesn't discount the fact that many people are building exactly that these days.
I'm tired of people on Hacker News making blanket arguments like "never do this" especially something as bland as requiring javascript. They have absolutely no clue what they are talking about.
Bottom line is if I (and Google and 37 signals and countless others) can choose to build something that doesn't support even IE8 I can quite happily choose to require Javascript to use my web application. I would be an idiot however if I required it for my blog.
http://developer.yahoo.com/blogs/ydn/many-users-javascript-d...
For clients in the UK, it's a legal thing:
https://www.gov.uk/definition-of-disability-under-equality-a...
http://www.rnib.org.uk/professionals/webaccessibility/lawsan...
FYI: Screen readers nowadays can run JS, but there are hiccups abound. Also I don't recall seeing "never do this", though I did see this: "JavaScript isn't evil by any means, and it's really important for creating cool shit, but it should never be required" from https://news.ycombinator.com/item?id=6176036 . Which, in context of the article, seems pretty reasonable to me.
This all started with when kintamanimatt mentioned that the article's technique breaks Progressive Enhancement, "JavaScript is useful, but nobody should assume it's present."
Obviously, I'll need to get an Xbox to play an Xbox game. I'll need Flash enabled on the browser to play a game built on that. Same goes for JS. But, again, the OP makes no mention of "applications" or "web sites" so I'm not sure why you're bringing up "the latest interactive experiences or web-based gaming" into this.
1. You're article actually states how small this is so point taken I guess
2. 99% of screen readers today actually support javascript [1] and as I stated, if you want to talk hiccups effectively supporting the visually impaired is a veritable minefield of challenges and just supporting javascript being disabled isn't even the tip of the iceberg.
3. Finally just this: Also I don't recall seeing "never do this", though I did see this: "JavaScript isn't evil by any means, and it's really important for creating cool shit, but it should never be required". Just read it back slowly, you'll find the word "never" in there if you're more careful.
You are also choosing to presume incompetence of these issues on the part of the author where I see none demonstrated. Even he bothered to explain his, fairly common these days, use case (JSON response processed on the client with javascript). If you want to argue against this sort of thing be my guest, but you're not really going to reverse the trend. Either way it isn't a debate I'm very interested in.
Reductio ad absurdum as they say.
[1]: http://www.punkchip.com/2011/03/why-support-javascript-disab...
That percent should be even lower now instead of being higher.
To start, it doesn't work if you use VPNs. Second, it doesn't update when you travel.
Here's one example. Even though I'm logged into Google and used Google Maps' 'pinpoint my location' feature, Google Reader is currently emailing me my daily schedule (a feature I enabled recently) in the afternoon. That's also despite my appointments being annotated with timezone information.
I call this bad engineering.
Also, it'd be nice to ALERT me when my timezone has changed and give me an option to keep the old one, or use what you think my current one is.
Personaly, I think for anything where timezone is relevant, the best move is to always let the user manually enter the timezone he/she wants, or have it match the timezone of the event/service he's looking for, and of course allow for override, so changing timezone has to done client side.
I.e. if I am in Germany and search from a flight taking off from Taipei two days from now, I don't care about the departure time from the German timezone. Same thing if I reserve a restaurant in Germany while I am in Taipei, you'll want the german time for the reservation. For something internationally broadcasted showing the UTC time as default and provide a way to adapt to other timezone is OK, as the event has no local anchoring anyway.
People can do the calculation for timezones, eventually some opt to not change their phones timezone to keep in sync with other more important schedules, etc. Trying to do 'smart' things seamlessly is often just screwing with the user.
tl;dr: the timezone you use depends not just on the user and their environment, but also the event and its duration.
"Luckily, the browser already knows what timezone the user is currently in, so we can make use of the getTimezoneOffset() function"
Getting the timezone offset isn't the same as getting the timezone. Granted that in certain cases having the offset is good enough, but sometimes you need the actual timezone. In my example I needed to know when a merchant wants their pre-order to be released. For that to work predictably I needed to know their exact timezone in order to adjust the time for daylight saving properly. Remember that there are quite a few timezones that are UTC-7 and they can behave differently in regards to daylight saving.
My laptop has never changed timezone automatically. And that is a feature in my eyes.
How can you assume that just because I'm doing a weekend trip that I want my whole life to be centered around the timezone in that particular place?
Of course, it can always be turned off for others, like yourself! :)
On a phone (which also acts as a wrist watch for most people), yes. Laptop, just no.
What OS does this? And how?
I think it's pretty safe to say most people want their laptop's clock to match their local time, because how is a wrong clock useful? Of course, if you're travelling for business with complicated schedules and meetings in home times, you clearly might have different needs, but most people just want their clocks to be right.
And you certainly don't want to keep your original timezone but change the clock setting -- then just everything's wrong! :)
And it is moronic for the OS to do this change behind my back (I don't know how OS X present this but thankfully Windows 7 doesn't do this madness by default).
How do I know that my OS got the time right? (did it perhaps get the time from my VPN location? was the network I connected to badly configured?) Since a laptop isn't connected to the net at all times I also have to consciously be aware of whether I've even been connected yet in this particular time zone.
"How do I know that my OS got the time right?" Well, have a look. I've often changed the DST switch manually after noting it not being correct. It wasn't then overridden by the OS "knowing better".
So, I have never ever (after the OS installation) changed the time zone for any of my computers and knowing that the OS doesn't do this behind my back I've never had to look or worry about it.
But even if it were perfect, why would you want your laptop to reflect the time zone you are currently in? Shouldn't the time on your laptop be the time you are the most familiar with and that you most deal with? For your day-to-day needs you have your phone and/or watch anyway, if anything a reference to the home time would be preferable instead of having all devices say the same thing.
This touches on to the relevant point of DST. I no longer have to worry about changing clocks; the worry has been replaced by the confusion over whether all my clocks have changed or none have.
At a bare minimum I'd expect any change of time zone to prompt the user.
Especially when you set the alarm for the next morning - and you know that DST kicks in but you have no way of knowing whether your phone knows that. And you can't change time beforehand because you don't know whether the phone will respect that or change the time back or make the change twice...
I found out that the stock android alarm app solves this (somewhat), when an alarm is set a notification tells you how many hours (and minutes) remain until the alarm goes off and this takes DST changes into account so you can see that it got it right.
Apple has succeeded in getting this wrong several years in row with the iPhone, quite embarrassing - and made much worse by the fact that you as a user can not prepare for it.
And for the record, OS X does it by sniffing for wifi access points. The particular access points available at any given location is a pretty reliable means of tracking location, assuming that there are any (if there are no access points, it can't find your location, and therefore won't touch your clock). The only exception are access points that themselves move, e.g. wifi on planes. I would assume that these access points are disregarded.
This is the same system that iOS uses to enhance the GPS (especially on iPod Touches and Wifi iPads, where you have no cell towers).
This is the part that scares me and has gotten me into trouble in the past. Basically it means I do not know if my clock is local time, home time, time of the last timezone I visited or something random and thus I cannot trust it.
This isn't just a hypothetical, but has bitten me in the past. Fortunately it lead to me showing up for my train an hour too early rather than an hour too late. But after that I never let my OS play around with my timezone settings.
Phone, yes.
Laptop, yes. Times are still stored internally, and the timezone represents a localization. And if I go across the world for the weekend I see that locally, it is UTC+10, while entries in a DB are timestamped accordingly. Then I return home and I can look at the clock in my menu bar and see my local UTC-8, and all the while entries in a DB are timestamped accordingly.
I'd love to hear your use case for "if location and correct timezone can be reliably determined, it still should not be adjusted".
Why do I say this? There are very few use cases I can think of - though I'm happy to be educated otherwise - for doing this differently. In addition, Outlook and Calendar on my Mac both pick up TZ changes and automatically re-render my calendar appropriately.
I really struggle to picture a use case of "I fly from my home in Seattle back to Australia, and yet I want the clock on my menu bar to display a wall-time 18 hours behind my current location, rather than displaying the current time where my eyeballs are at, and changing back upon return".
Actually, I can think of one - the fear that poorly programmed or tested applications may handle this incorrectly. In which case I can understand this desire, even as a facet of "solving the wrong problem"/shooting the messenger.
Would these be the same studies that Microsoft cites to justify force-feeding Metro to desktop users?
You can reliably tell where a cell phone is by the tower it connects to, guessing by IP is nowhere near as accurate.
Garry Tan, cofounder of posterous, said of it : "Finally, timezones in javascript done right. The world has been made a better place via this fine javascript library."
Incorrect (though usually true, I'll grant). I've often seen people with older Windows laptops (newer Windows variants, 7 and 8 certianly Vista maybe, are much better at this) that are set to display UK time but think that they are in UTC-0700. This can be fun when using file synchronisation tools... If the OS doesn't know its timezone then the browser won't either. This is a moot point when you are talking to just one user, until that user fixes their clock settings of course, but becomes a significant issue as soon as you need to coordinate timed actions between users.
Detecting the local time offset from your reference clock isn't always useful either: the machine's clock could be completely wrong so it says 01:23 where the user's wall-clock says 21:00 and your reference time matches neither.
Also some users may wish to fix your interface to a timezone other than the one currently set on the device they are currently using. Someone on holiday or on a business trip might want to keep some things linked to local time back home so they find it easier to coordinate with people there. Conversely someone currently in their home timezone might want to force your interface to match a non-local timezone for coordinating with teams/friends elsewhere in the world. You can't rely on them adjusting their clock in this circumstance as they may not be using their own machine and so might not be able to adjust the time/timezone settings (or may wish to keep them accurate to current local for other reasons).
tl;dr: Dealing with time values when there are any international concerns it harder than we tend to assume.
And depending on your app, users might want control over the timezone. When a teacher creates an assignment in canvas-lms, the due date defaults to midnight of the selected day. It would be surprising if it set it to 2am, just because the teacher happened to be traveling when the assignment was created.
To fix this I run a cron job on the ST/DT boundaries that increments or decrements stored date/times in the event calendar.
Zip codes are based on postal routes, cities are not.
For example ZIP code 42223 spans Christian KY and Montgomery TN, and ZIP code 97635 spans Lake OR and Modoc CA.
Zip codes are also non-contiguous in some areas, and not bounded-polygons in others which makes them really bad for geo purposes.
You can't infer (reliably) your exact location from a zipcode because of the reasons listed in the other comments linked to your parent comment.
When it comes to Rotten Tomatoes, I'm glad it asks my location because I prefer US film reviewers to my local film reviewers. If I let Rotten Tomatoes know my time zone, I can't access US reviewers (that I'm aware of).
Theoretically, yes. In practice, you should just use UTC.
>If you're in a corporate setting and the majority of your users are in the same time zone as your servers, then it makes sense to use that time zone.
No it doesn't. Use UTC. The Internet does not have a time zone.
>There's fewer date conversions that way.
Your library is probably "converting" the date anyway, UTC or not.
>for the other half it's a 1 hour DST conversion - which many time handling libraries have optimized
Is it really that slow to convert a date?
If you need to pick a timezone which does not observe DST, that rules out (generally speaking) most of North America and Europe and a lot of other places. So maybe you want your new time base to be Asia/Tokyo, Asia/Shanghai, Asia/Kolkata, Europe/Moscow? You're still at the whim of governments who constantly tinker with DST -- sure Europe/Moscow is "fixed" now, but it wasn't as close as 2 years ago.
To each their own, but I would not get into the business of depending on any particular zone in tzdata staying as it is currently defined for any amount of time in the future.
Many time handling libraries are extremely efficient at adding/subtracting an hour from time. This is important because we have many legacy systems using this data (think mainframe COBOL to today's latest technologies with a smattering of everything in between). So it's not just a single time handling library in use.