Just to clarify: If you think you need a client because you are currently using an XMPP client with your gmail account, then you do NOT need to switch. See bullet point 1 — XMPP _clients_ are still supported.
Google enthusiastically kills legacy and niche products. I'm making wild guesses, but I don't think we'll have support for XMPP clients in the long run.
They're saying XMPP clients still works with the App Engine XMPP service (which is "An App Engine application can send and receive chat messages to and from any XMPP-compatible chat messaging service, such as Google Talk"). The email says that Google's new service no longer support XMPP.
That extension is essentially broken (running iceWM), as the text input box dips below the taskbar. Furthermore, it requires chrome to be open. I don't necessarily want to keep my browser open at all times, nor do I care to use my browser as a chat client. (Also, what happens if switch to some non-standard browser? There is no way that my choice of browser should impact my ability to chat.)
I'm trying to stay open-minded, but so far I really dislike this product. I hope gtalk stays open.
I think the Chrome app is an excellent cross platform solution (written once, easily maintainable). As for the protocol I don't think it is practical or even possible to built this sort of all encompassing communications platform on top of XMPP (see the competition).
I think the chrome app is one aspect of a cross-platform solution. But what if I'd like to chat without opening chrome?
I'll cross my fingers for a Hangouts API that allows me to write a standalone client. However, if chatting through a semi-broken browser extension (or memory intensive gmail tab) are my only options, then I guess the service isn't for me. I think the reason I'm annoyed is that I have a lot of important contacts on gtalk, so migrating away will be a challenge.
I know change is inevitable, that the service is free, and that they want to offer a better product. Perhaps my use cases are just too expensive to support in the long run.