Hm, this reply is kinda out of order... On Sun, Jul 24, 2005 at 01:14:11PM +0200, martin f krafft said > also sprach Rob Weir <[EMAIL PROTECTED]> [2005.07.23.1530 +0200]: > > Why are you trying to specify a name for it? Bazaar doesn't really do > > this anymore; instead, you have multiple locations for the one name, and > > bazaar figures out which to use based on context. > > I am not sure I follow, but it sounds cool. So let's take the > classic case: I have an archive accessible via read-only WebDAV as > well as sftp. Previously, I would register the first with an "-anon" > suffix, and the second normally. This would allow me to commit to an > archive from the same machine where I could also use cron to sync > automatically (using password-less WebDAV).
"baz archive-mirror http://arch.madduck.net/~madduck/pub/debianbook/"; should mirror *to* the http server, but I think I'm misunderstanding your use-case. > Since I cannot specify a suffix anymore and am thus bound to using > the official name of the archive, I try to register both > successively, but I find that they overwrite each other: > > cirrus:~> baz --version > baz Bazaar version 1.4.1 ([EMAIL > PROTECTED]/dists--bazaar--1.4[bazaar--devo.cfg] > [...] > cirrus:~> baz register-archive -d [EMAIL PROTECTED] [364] > cirrus:~> baz register-archive > http://arch.madduck.net/~madduck/pub/debianbook/ [365] > Registering Archive: [EMAIL PROTECTED] > cirrus:~> baz register-archive sftp://[EMAIL > PROTECTED]/srv/arch/madduck/pub/debianbook > Registering Archive: [EMAIL PROTECTED] > cirrus:~> baz archives | grep -A2 [EMAIL PROTECTED] [367] > [EMAIL PROTECTED] > http://arch.madduck.net/~madduck/pub/debianbook/ > [EMAIL PROTECTED] What does "baz whereis-archive --all-locations [EMAIL PROTECTED]" print? "baz archives" now only prints one of the registrations of an archive, apparently the primary (aka master). If you look in ~/.arch-params/archives/, you can see the data that's stored directly. > This is a real shame, because it effectively loses the extra layer > of abstraction of the archive name. I really enjoyed being able to > register an archive as "zope-upstream-archive-anon" and just use > that name instead of the [EMAIL PROTECTED] crap. The aim of the move is to let you use urls instead of archive names, preparing for the day that names are optional (or at least way less obvious). Have you run "baz upgrade" yet? It will convert your existing archive to the new registration system, then you can do things like "baz get http://arch.madduck.net/~madduck/pub/debianbook/"; to check out from the webserver copy of it... > Why do you require the archive name I use on my work machine to be > bound to the upstream name? Because they're nominally the same archive, just different (possibly incomplete) copies of it. You can mostly not use archive names at all now, and just refer to things by URL. -rob -- Words of the day: strategic Skipjack Elvis industrial intelligence credit card -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
