Is "bzr push" safe?

[Andrew Cowie]

On Wed, 2011-11-30 at 12:14 +0900, Stephen J. Turnbull wrote:
> This can be disconcerting in any VCS, but especially so
> in bzr which "hides" dotted revisions by default (A and B can actually
> be long sequences of revisions, including merges -- if what developers
> perceive as the "mainline" flops and becomes dotted revisions, it
> seems like a lot of work "just disappears" -- panic city!)

A good example of why Bazaar `log`'s default of suppressing revisions
that don't happen to be on the left hand edge of the graph is such a
poor choice of behaviour. 


> but you *also force renumbering of
> existing revisions* 

[what you perceived as the numbers of your] existing revisions... of
course, someone extremely experienced in Bazaar would know the unique
revid for the revision hasn't changed, but still, you don't learn that
until you've stuck with it for a long time through sever propose /
accepted / merged / pulled / review cycles.

Personally, I don't think anyone would care about the apparent revno
renumbering if there was some sense that 'mainline' was "stronger" than
other revisions, like, say, bold faced. Even if mainline wasn't on the
left hand edge of the graph (as it so frequently is not in anything
other than a mirror copy of it) you could still see if threading its way
through the graph. That would be really cool in `visualize` and I think
would help `log` too.

The trick would be how to bless such revisions. Some signed-of by tag
[equivalents, perhaps]. Or an attribute set when it's pushed into a
named ("mainline" ?) repository (via hook?) Or...?

AfC
Sydney

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 836 bytes
Desc: This is a digitally signed message part
URL: <https://lists.ubuntu.com/archives/bazaar/attachments/20111130/3f765cbe/attachment.pgp>