bzr-specific protocols

Aaron Bentley aaron.bentley at utoronto.ca
Wed Aug 31 16:23:42 BST 2005


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

I'd like to propose that we leave the protocols the way they are.  It
harms interoperability to use a specialize protocol when in fact, these
are standard protocols.

I like the fact that I can click a bzr link in my web browser, and
explore the working tree.

I think it's good that my working tree might get indexed by Google.

When we have sftp support or ftp support, we can use standard mirroring
utilities to copy branches.

It will lower the barrier to entry, because people will see http: and
say "hey, I know that.  Now I can write a little bzr add-on using
urlgrabber".

I don't think the argument about multiprotocol tools holds much water;
multiprotocol tools can determine that this is a bzr tree the same way
bzr does: by testing for the presence of specific files.  I do this in
Bugs Everywhere, for example.

Perhaps we could use the fragment identifier to hint the resource type:
http://panoramicfeedback.com/bzr.24#bzr

This seems appropriate, since it is a client-side hint, bzr being a
particular 'view' of the data.  It would be handy when the specified
resource is versioned using several systems, e.g.
ls -a foo
{arch}
.bzr
.svn
CVS
README

Aaron

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.1 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org

iD8DBQFDFct+0F+nu1YWqI0RAiMfAJ4462aGehcTMbrU9UVHy8VEyzoyzACffz10
+9lW57NJKtYUi0GGLUlt0MQ=
=ZGhG
-----END PGP SIGNATURE-----




More information about the bazaar mailing list