I cannot count the Tuesday nights I spent banging my head against the wall, just to find out that Facebook broke something again. It's one of the many prices you pay for being a slave to The Platform.
Hate to say it, but this is par for the course every time Facebook releases new APIs. They move at an insanely fast pace for a company of that size, and there's no free lunch, so the tradeoff is buggy software.
I don't see a lot of first mover advantage with the new stuff, so my advice would be to wait a few weeks.
I'm definitely not criticizing the developers. Developers are often pressured to release half-baked features for the sake of a PR push on a particular date.
However, I don't believe a company should be able to enjoy massive PR attention unless they've also delivered the product they've claimed.
As developers, we also shouldn't sell our own time short. Based on the comments here, the low quality of this release has cost several of us some hours/frustration hacking away on a broken platform.
If you have already working app, you can wait a few weeks. But I'm creating a new one - I have no desire to write it using the old libraries now just to rewrite it later.
For me (so far) everything was working except that they haven't implemented some very needed features from the old FB JS lib in the new one :(
I spent a good portion of Friday trying to get their new JS libs working; no dice. Even just running the three-line code examples from the docs was throwing errors deep within the compressed JS. I had great results with the OAuth2 API, however.
This is par for the course on a daily basis. I work on facebook apps essentially all day, every day. One of the biggest challenges is writing an application that relies on facebook functionality that may or may not vanish or fail hourly.
+1 for building a business on a platform that you don't own, don't control, and has no TOS or SLA with you that protects your uptime.
This isn't exactly news at this point, but I was also having this problem just today and was whining about it and here is this post. It got to the point where I was figuring out what functions actually existed by using Chrome developer tools and investigating the FB object and its members, and then trying to read the minified javascript.
On the plus side, it's bound to improve soon for two reasons:
1) It can't get any worse
2) They can't get away with this much longer
I happen to kind of know one of the guys responsible for a lot of the JS API stuff, so I plan on giving him crap about this next time I see him, if that's any consolation. :-)
Usually this is the way most of "take over the web" revolutions appeared
to be after the buzz is down.
(e.g. Have you seen any interesting Google Wave plugin recently?)
I had basically the same experience updating cornify.com to the new social plug-ins. Documentation is incomplete, the Facepile doesn't work at all, and using an older Facebook Connect library at the same time breaks everything.
This has been my experience for a long time now. Even when talking to big clients at work, we're simply not encouraging integration with the Facebook APIs since we can't be sure what's going to happen.
Another thing that is a big downer is that the plug-ins can't be skinned. This may be good to keep the brand visible and establish a minimum level of design quality, but for agencies it doesn't work since it's impossible to create something dead-sexy.
> One peeve: an "undocumented feature" is not a feature that was advertised but doesn't work, it's one that wasn't advertised but does work.
"Documented" simply means "has documentation written for it". Being documented or not is independent of whether a feature (1) works or (2) was advertised.
Where the author wrote "In some of the talks they admitted that although these features are live right now, they are not yet documented (ie. you can’t use them)", I think he meant that some features were announced, but since they are currently undocumented they cannot be used, simply because you can't figure out how they are used. (And they might or might not work.)
For a couple of days, new Like button stuck showing only 5 likes for my site whereas I'm sure there are more, so I removed it.
But, implemented a Facebook login though the new graph api and it's really nice to work with. The documentation lacks a bit but enough to integrate it to my site in a day.
It drove me insane that my app would be horribly broken and all I could do was wait around spamming refresh.
Seriously awful experience, and it sounds like they're continuing the tradition!