This was the only problem. If firefox broke our "browser detection" we couldnt continue doing it that way. It was not safe to keep it like that.
The userAgent is the only one thing that will not change just because if the browser vendors change it's structure they will break thousands of site all over the web. And they don't want that. You should *avoid* the use of browser detection as much as you can and use feature detection for you specific problem/bug instead. If you can't do feature detection for your problem/bug, after googling about it for at least 10 minutes (this is a rule :D), than use browser detection. Hope it helped. -- Fábio Miranda Costa front...@globocore *github:* fabiomcosta *twitter:* @fabiomiranda *ramal:* 6410 On Wed, Oct 13, 2010 at 9:19 AM, Olivier El Mekki <[email protected]>wrote: > On 00:23 Wed 13 Oct , Ryan Florence wrote: > > If you are browsing the web with a faked user agent string, that's your > fault. > > Just like it's your fault if you disable javascript. Still, it is > considered as a security good pratice. Even if a very few people are > doing it, we can't blame them for trying to be secure. > > Of course, it is not as pragmatic as nojs support, since this last one > is also important for search engine index. > > We could merge the two methods : checking the user agent, then matching > it against feature detection, setting some 'double_checked' flag that > let mootools user be sure about the browser, and not implementing crashy > feature if not. > > Or we could use feature detection and use the user agent in case feature > detection isn't enough. > > I suppose you debated that when switching browser recognition method. > What were the problems of feature detection, except the firefox broken > one? > > -- > Olivier El Mekki. >
