News for hackers. Stuff that matters.
Re:Most expensive javascript ever by gaius on Monday July 20, 2009 @12:38PM (#714654)
Well, for a start, it would mean rolling out Opera to every developer, tester and support rep's desktop, and it being someone's job to make sure it worked on every PC they had, to keep it up to date and apply any patches. It also means doing any manual QA cycles again with Opera, and/or rewriting automated tests to use it too. All of those things have cash costs associated with them.
Re:Most expensive javascript ever by tokenadult on Monday July 20, 2009 @01:25PM (#714734)
Well, for a start, it would mean rolling out Opera to every developer, tester and support rep's desktop, and it being someone's job to make sure it worked on every PC they had, to keep it up to date and apply any patches.

Is it impossible to write a Web application that simply doesn't test the limits of most browser implementations? Again, I'm asking an informational question to follow up on your interesting response. What actually happens in the typical case of a server vendor's application that is supposed to run on an arbitrary browser if a browser the vendor hasn't considered is used? Does the whole application crash? Or does it just look a little bit less pretty?

Re:Most expensive javascript ever by akeefer on Monday July 20, 2009 @02:40PM (#714871)
It could be practically anything; sometimes it's a little less pretty, sometimes it's a lot less pretty, sometimes a few little parts of the UI don't work as intended (maybe changing field X doesn't update field Y like it should, maybe button Z doesn't do what it should when you click it), and sometimes it just starts throwing exceptions like crazy and completely blows up. CSS is kind of sort of standardized, so you sort of have some hope there that all standards compliant browsers (not that any of them really are 100% compliant) will render the page about the same way, but once you get into Javascript land the myth of standardization quickly disappears, and multi-browser support becomes a never-ending nightmare.

Unfortunately, you don't know which of those end results you're going to get until you actually try the thing out, and the answer could be wildly different on different pages of your app.

It's still generally poor form to straight up disable a particular browser unless you know for a fact that it doesn't work properly (generally in the sense of you-can't-use-the-site, not things-look-kind-of-funny), but perhaps in this case they did in fact know that their app didn't work with Opera.