New to Hg - best practices?

[Marcos Scriven]

Hi all

I've tried out Mercurial a bit, followed the tutorials, read the first  
few chapters of the book, and read bits of the wiki, including the  
'best practices' page.

However, as a long term CVS user, and not exceptionally bright Java  
developer, I'm still left a little dazed as to how I would best make  
use of it.

I should say that my interest in Hg is as a lone developer. I like its  
speed, the ability to not need a server, and to try out things and  
commit without needing to mess up my integration repo. I don't have  
delusions of grandeur that my 'Hello World' app is going to attract  
hundreds of developers accross the globe.

So - to the questions:

1) I get that I should (could) have one 'in' repo clone, one 'out'  
repo clone, and one or more working clones. What I'm not sure is how  
to integrate that with an IDE? Do I effectively need a separate IDE  
project for each working set? Do I have to blow away that project when  
I've finished with that working set?

2) I know it's best to get away from having a central repo, but  
presumably one does still need at least one repo to be 'the' repo, or  
master so to speak? How do you guys manage this? I understand you  
could have one repo for different versions of a piece of software, but  
not sure what's the best way to handle this.

3) I am writing software that has a server, and multiple  
implementations of clients. Therefore client version x would be  
compatible with server version a,b, or c. I realize this isn't  
necessarily a source control issue, but wondered if you had any Hg- 
specific pointers that might help to manage such depedencies, at least  
in simple terms like 'I have cloned the source for server version x,  
now give me a clone of the client compatible with this'

Thanks, and appologies if my questions are a bit noddy.

Marcos