When announcing this package release, Intel's Gordon Jin stated "We are moving to UXA/DRI2/KMS, so this is probably the last release supporting EXA/DRI1." DRI2 is great and certainly superior to the outdated DRI1, so that point is moot unless you want to build xf86-video-intel 2.8+ on a system running old packages. On the EXA side, if this is indeed the last quarter where the Intel X.Org driver will support this standard 2D acceleration method, that could cause a few problems.
UXA was conceived to sort out pixmap memory management in the world of 2D acceleration and be the grounds for testing out other changes when using a kernel-based memory manager (such as GEM). UXA is based off of the EXA API and for a while they planned to merge the changes back into a new version of EXA, but that idea was scrapped.
UXA does offer better performance, but at this current time there are some very prominent bugs outstanding. We have talked about these problems before, but they largely deal with stability issues and graphics rendering defects. The UXA problems can be common among users so much so that Canonical decided to disable UXA and continue using EXA as the default acceleration method for Ubuntu 9.04. For those users not running into such bugs, UXA is faster though in some areas there are still performance regressions where EXA does better.
In order to use UXA, your Linux kernel must support the Graphics Execution Manager, which went mainline only in the Linux 2.6.28 series. Therefore, if the EXA support is indeed stripped out of future Intel driver releases, but you are still running an older kernel, you will need to upgrade your kernel if you wish to upgrade your Intel driver and still have 2D acceleration. UXA or having no 2D GPU acceleration would be your options.
Let's see how this evolves and whether the next quarterly update due out around July has EXA support (whether it may be stripped out completely or limited to a build-time argument) and whether Intel has worked out its issues.