dzsekijo at creo.hu
Wed Mar 30 22:32:56 UTC 2005
When I first looked at "alternative" VCS software, I preferred the arch
way over the darcs way in respect of the notion of a repo.
I thought that the software saves me headache by maintaining a registry
of known repos, and I don't have to care about where my scratch/working
copies are, I can always use a global, protocol-neutral namespace to
locate what I'm looking for (and then enter a
"checkout/hack/commit/forget working copy" session).
However, later on I realized that I -- and guess most of us -- don't
organize my work as illustrated above. Rather I have a permanent working
copy and that's what I look up first when I wanna do some work. Thus the
repo registry rather seems to be a constrained and ad-hoc mapping
beetwen the filesystem namespace (which tends to be the primary
reference) and the artificial repo namespace. This consideration (among
others) made me prefer darcs. Hello darcs.
However, again, what's in the above paragraph is not 100% true. That is,
it is true when I working locally. But when I want to commit (push) to a
remote repo, the url and the user at host:dir naming schemes are ad hoc,
again. When I work locally, it turns out immediately by interactive
shell usage (powered by readline) whether I keep my projects under
~/projects, ~/projects_2005 or ~/projects-2005. But I might like not to
start an interactive session on the remote machine, or look up the exact
url again; and anyway, I'm likely more familiar with my local filesystem
layout than a remote one or with the dusty corners of the web.
So, what do you think of supporting a lightweight, non-constraining form
of archive registry, namely bookmarks?
You could just do "darcs bookmark <repo loc> <nick>", "darcs bookmark
-l" to list bookmarks (and have some other options, like deletion), and
you could use the nick as reference being equivalent to the full location
during ordinary interaction.
More information about the darcs-users