Ouch.
I first met Clearcase in a solely Unix
environment. With a couple of config files to set up views everything ran
perfectly and seamlessly off the command line. It worked by using symbolic links
for all unchanged files to the repository somehow (it was 6 years ago and I
never really got into how it worked). If you changed a file you simply removed
the symlink and copied the file into your working directory to edit
it.
Since using Clearcase I have met a few
different versions of version control systems, but Clearcase remains one of my
favourites. We used it to have different branches for slightly different
products. We set up views to pull common functionality from a single branch, and
then all other code came from the relevant product branch if it differed from
the core code but from the core code otherwise. You could set up wildcards int
he branch specification. It was really cool.
Anyway, the argument that it does not
integrate well with Unix is not valid. It may be a bit more tricky to set up,
but it does work really well.
Alastair.