The first is that for the last 2 years WWDC has been all about iOS. So either (a) there is a return of focus in WWDC to OS X or
It isn't so much lack of focus but lack of anything new to talk about. There was nothing new to talk about because Snow Leopard had been released ( and covered heavily at previous WWDC) and nothing new had been released. They could have had some "real work usage and lessons learned from SL" sessions but given all the new stuff had to cover with iOS can see how those 'rearview miror' sessions easily got squeezed out.
(b) early next year they'll announce a brand new OS X focused conference.
What? They are probably announcing the new OS at this event! The next WWDC will be the first time can have classes on the features and API mutations they announce here. Given many developers have their hands full with development of their own products, many won't time till next June to take several days off to play around with 10.7 features (or will have played around and know need help with some of the more tricky bits. ) To have effective classes on a topic in June need to give folks the information now so that they can planned to be there and know why should.
The second issue is that it seems kind of odd to release a brand new OS as holiday time approaches.
10.3 Panther released Oct 24, 2003 . Didn't really kill its adoption.
10.1 Puma Sept 25, 2001 (which is technically Fall )
Frankly, a 15 month beta cycle would be safer with a complicated stack as a OS deployed on a wide spread of hardware and a deep software legacy portfolio . OS X releases often break older hardware drivers and app software. Remember Mac OS X Server is also bundled in this release. People with machines which run their business don't want an OS that hasn't been thoroughly tested. I also suspect that once give the OS to many users in the beta cycle that they will find tons of bugs that need fixing that Apple didn't catch. That
always happens.
Could easily see this this as a January 2012 release date. However, Oct/Nov isn't bad. It is an "upgrade" for users. If Apple has moved thei bulk of their hardware updates out of the Oct-December time frame then no "need" for an OS update in that window. However, if Apple's QA process is extremely better than industry average and 10.7 is somewhat conservative on changes then they could hit a late Sept deadline. I doubt that happens because don't think Apple's processes are dramatically better than everyone else.
As for features of 10.7, who knows (only

of course). However, I would hope that with Snow Leopard's focus on the under the hood type stuff, we'd see more really nice end-user type features show up. Maybe this is where we finally see the "marble" interface?!?
I'd rather Apple add real updates to the OS rather than tweaking superficial eye candy.
Maybe, this is where they bring ZFS support?!?
Given Oracle is trying to sue Google over Andriod for megabucks ... that's doubtful. Oracle probably wants significant money for Apple to license ZFS. I doubt Apple would pay. ZFS made much more sense when Apple had large storage products. Now they don't. ZFS updates are going to come, ,but the will come first in the "closed" Solaris version. So if Apple wants early access to the updates they'll need to pay. I doubt Apple will restart ZFS after putting it on the shelf while its future was cloudy.
More likely Apple would put Filesystem developer resources into TRIM support and minor, more API transparent, changes to HFS+ . It is not absolutely necessary, but it is useful. SSD/HDD file caching would be another bang for the buck without really changing the API all that much. [ namely move the HDD hot spots to a SSD drive. Kind of like building a hybrid drive with discrete components. there are enough 2+ drive unit macs that this doesn't have to be a "rarely used" feature. ]
I don't think Apple is going to do anything to dramatically improve data integrity along the lines that ZFS does.
the much
bigger gap is not having generic USB 3.0 support. I can see them limping along in 10.6 with custom driver only support for any hardware they ship in the interim.
There a several data security updates ( self encyrpting disks , x86 on chip AES support , etc. ) that should get baked into a 10.7. We'll see if that is "too boring" them to pursue.
I'd also expect some enhanced support on CPU+GPU systems also. ( better and/or transparent OS leveraging of "extra" GPU for OpenCL work )
Also a better integrated/discrete GPU automatic toggling. The system they have now is quirky. The discrete GPU triggers for some users triggers most of the time.