server less collaborative project

[Sciss]

hi,

i'm new to mercurial. i have a mental problem understanding the  
architecture... in our case we want to use mercurial for a  
collaborative project between four persons. we do not have a server  
that can act as a central mercurial repository, actually that is why  
i began to read about mercurial, as i'm already pretty fluent in SVN.  
but although it is stated everywhere that mercurial is  
"decentralized", in practically all tutorials and wikis people seem  
to assume that there _is_ a server that hosts the repository.

well, i tried to set up the scenario without the server, but found it  
very tough and unergonomic. here is my approach:

user A (me) creates the initial repository:

A$ mkdir mercurial-repo
A$ cd mercurial-repo
A$ hg init
A$ hg clone . ../mercurial-work
A$ cd ../mercurial-work
A$ hg pull
A$ hg up

... ok, i will place some file here

A$ cp someOtherPlaces/someBasicInitialFile .
A$ hg addremove
A$ hg commit
A$ hg bundle --base null ../initialFullRepo.hg
A$ hg push	// otherwise the repo is out of sync
A$ cd ../mercurial-repo
A$ hg up	// otherwise the repo won't accept further unbundles...?

----> sending initialFullRepo.hg to my mates via email. So for  
example person B will receive that file and do on their own computer:

B$ mkdir mercurial-repo
B$ cd mercurial-repo
B$ hg init
B$ hg unbundle ../initialFullRepo.hg
B$ hg up
B$ hg clone . ../mercurial-work
B$ cd ../mercurial-work
B$ hg pull
B$ hg up

now B works on some files...

B$ vim someBasicInitialFile
B$ hg commit
B$ hg bundle ../changes2.hg	// must be before push, otherwise it  
fails with an error?!
B$ hg push	// otherwise the repo is out of sync
B$ cd ../mercurial-repo
B$ hg up 	// otherwise the repo won't accept further unbundles...?

now B sends with file changes2.hg to the mailing list. Now i (A) will  
update like this:

A$ cd mercurial-repo
A$ hg unbundle ../changes2.hg
A$ hg up
A$ cd ../mercurial-work
A$ hg pull
A$ hg up


i find this very complicated and error prone -- six lines of code to  
update my copy, five lines of code to commit something, and when you  
forget one of them, you have to fiddle around a lot). i don't  
understand why i am forced to
- have a separate working dir
- unbundle into the repo folder (instead of the working dir)
- needing to call "hg up" after the unbundle in the repo dir
- needing to bundle before calling "push" in the working dir

isn't there an easier approach for dealing with our situation? Maybe  
i'm missing something? or should i just write two shell scripts that  
collapse all the above commands into a "easyUpdate.sh" and  
"easyCommit.sh" that make the thing work more like SVN?

thanks for help! ciao, -sciss-