All the protocols were open-sourced as part of the initiative. I think they envisioned a world in which inside-the-wall installations existed in the same way the companies have Exchange servers today.
Except Microsoft is willing to sell you Exchange Server but Google wasn't willing to sell you a Wave server. Writing your own is very difficult; no one has managed to match Google's implementation.
Do you know what the problems are that make Wave servers hard to write? I'm writing something similar right now, and if there are enormous problems coming up, it would be nice to know about them now.
The real-time UI appears to be the hardest part; Google's client is something like 20K lines of GWT code. (Technically the client isn't part of the Wave "server", but since AFAIK you aren't allowed to use Google's client with your own server, if you write a server you also have to write a client. Also, I don't think the distinction is particularly relevant in Web apps.)
Absolutely. Wave looked like the perfect way for businesses to run project collaboration, but I can't use it at all until its inside our firewall and we control the data.
Sure we could spend time and effort to re-create the wheel, but if Google wants to monetize it, they will do the work for us and provide us with a product we can purchase.
One wonders how the folks on the Novell Pulse team are feeling about this announcement. And how many hours they may have spent on "interoperability with Google Wave through the federation protocol".
There's no reason they couldn't provide an appliance (like the Google Search Appliance: http://www.google.com/enterprise/search/gsa.html) or simply make it available for a local install (like every mail server used by every company that cares about things being 'inside the firewall').