[Gnu-arch-users] bitkeeper vs tla

[Zenaan Harkness]

I finally realised the primary user (/developer) level structural
difference between bitkeeper and tla:


<lights=blazing! users="bitkeeper refugee migrants">

        in bitkeeper, a tree is an archive is a tree;
        in tla "archives" and "working trees" are different entities

</>


This impacts different things that you want to do such as share an
archive, merging with a peer/parent, etc.

Even though I'd only trialled bitkeeper for about two weeks, not being
particularly congnizant of that distinction was a source of confusion
for me up until now, when it finally clicked.

I don't yet have an opinion of which is preferable. I think there are
probably facets of both that are appealing (I speak from a user's
perspective, not an implementation perspective):

I like arch's "archives" formalism.

Bitkeeper is simpler conceptually (for the user) which aids newbies.

In bitkeeper, due to this lack of formalism, archives at least appear to
be more "lightweight" entities:

* one can easily end up with a tangle of "parent-child" relationships

* "bk clone parent-dir new-dir" makes a new tree (which, now that I've
been pointed at shell variables, is all but the same as tla get -
perhaps the name of the archive could be embedded in the archive, so one
must only point to the archive url to "get" a tree, and behind the
scenes, it's could be automatically registered, revlib-cached, etc)

* "bk push" and "bk pull" seem to be the equivalent of star-merge
inwards and outwards - again, they are short and therefore to the naive
user, "lightweight" commands that Just Work.

cheers
zen

-- 
Homepage: http://www.soulsound.net/
Please respect the confidentiality of this email as sensibly warranted.