Some unscientific timing results (on the Python source tree)

[Paul Moore]

On 31/03/2008, Ian Clatworthy <ian.clatworthy at internode.on.net> wrote:
> As I've just mentioned in another thread, repositories converted via
> bzr-svn use a different scheme for revision ids to what Bazaar does by
> default, leading to a different sized repo than that converted by other
> means. I'm yet to measure the difference myself so I'd be interested in
> seeing some numbers if anyone has them.

I've no useful numbers to add, I'm afraid, but I do find it somewhat
worrying that it's possible to get significantly different performance
and size results depending on what choices you make when creating a
branch.

I know it's a matter of trade-offs, and I can see that having choices
is not in itself a bad thing, but to be honest, it seems to me that
Bazaar gives me too many choices, and not enough information on the
trade-offs (quite possibly because it's not possible to give that
information - "it depends" on too many factors like repository size,
network speeds, what operations are likely to be common, etc etc etc).

Is there any plan to reduce the number of options for how to create
branches in Bazaar? I suspect not, as this seems to me to be a
fundamental part of the way Bazaar supports the various workflows.

Paul.