[Gnu-arch-users] Arch vs. Darcs/BitKeeper: WC & Respo vs. WC == Repo?

[Deliverable Mail]

I was using BK for a long time, and wanted to come clean, become a
good FLOSS citizen and use darcs or arch...

Nah.  :)  BK suddenly hiccuped on windows -- I fixed it, it really
needed ssh-agent and seemless login -- bit otherwise nothing beats its
speed and help system...  WIndows-linux interoperability is amazing
for bk.  Darcs cares about it, to, UNison does it, SVN does it, and
any SCM which want to take root wants to do it.

But meanwhile, for the few days it took me to fix BK, I tried arch and
darcs.  To the point where I spent these two days compiling tla on
MSFU and getting it right.

Now I'm a bit amazed as to why people bother with those interminable
names in tla?  In star-merge, you have to say things like tla
star-merge <email>--<archname>/c--b--v which takes forever.  The thing
gets stuff into dirs with suffixes like --base-0 which I go and cut
right away, as it doesn't stay that for long, becoming a --patch-#! 
And naming, do you really have to fix all the regexps before tla will
commit? For whatever reason, setting untagged-source junk didn't work
for me -- all those untagged files still were reported by tree-lint as
looking like source but not being it!

Overall, in a distributed mode, you'd have to keep a copy of merge-dir
WC from a remote archive, and a local tagged archive and _its_ WC, and
merge from the latter WC into the former and then commit into the
remote archive to sync.

Now, if I use a laptop and simply want two full repos with all
histories on both and sync them when I get a chance to, e.g. if I work
on a big screen at home and on my laptop in a cafe, -- why do I need
the second layer?  What is it that tla enables through such two layers
that darcs and bk don't?

Cheers,
Alexy