The importance of patch pools

[Kevin Smith]

Greetings all,

I have been on a quest for a distributed SCM for well over a year now, 
and am (again) exploring bzr. Three features are particularly important 
to me, and I haven't yet found a system that provides all three:

1. simplicity
2. plain http hosting
3. full cross-platform functionality

Bzr looks great for #1, and I believe it works fine for #2. My questions 
have to do with #3. Although I only use Linux myself, I work mostly on 
teams where other developers use Mac and/or MS Windows boxes.

It should be pretty easy to get bzr to run under MS Windows. But on a 
typical Windows box, hardlinks are not available, so if you have to copy 
an entire tree to create a branch, branches aren't cheap. If branches 
aren't cheap, you have to change the way you work with the distributed 
SCM tool. And if you can't use all the functionality of the tool on 
other platforms, it probably won't work for the teams I'm on.

I know there have been mentions in the past of creating patch pools. 
When patches were identified by hash, and never modified, that seemed 
pretty straightforward. But if bzr is going to store revisions as 
weaves, where does that leave patch pooling? Will it still be possible?

Thanks,

Kevin