Issues with bazaar 1.1

[James Westby]

Hi,

Sorry for the delay in replying.

On Thu, 2008-01-17 at 15:05 +0100, Ingamells, David wrote: 
> I have recently started using bazaar on a Ubuntu Gutsy platform and
> have encountered some problems:
> 
> 1) I am using the Ubuntu Synaptic package manager in the way advised
> in the bzr installation instructions for Ubuntu. I am not allowed to
> install bzr-gtk (the latest available is 0.90) with bzr 1.x. It
> complains that bzr-gtk requires bzr < 0.91~.
> 
Hopefully this will be updated soon.

> 2) I have the following issues with bzr 1.1.
> 
> I am trying to write some tooling that uses the bzr command to manage
> a task (aka change set) based system based on bzr's revisions with a
> central "master" repository. Before allowing a user to perform a
> commit with this tooling I want to do some quality checks to ensure
> that the update builds with the latest updates from other developers
> that have been already merged into the master repository. To do this I
> need to run some bzr command that gives back a status to my script
> about missing revisions and merge conflicts. I have been trying
> several commands and find the following:
> 
> a) "bzr missing --quiet" is not quiet. It always prints a message even
> with no missing revisions
> 
> b) The returning of a useful completion status is patchy with many
> commands:

> 
> The documentation is very scant about saying when and what return
> statuses are used.
> 
It would be great if you could file bugs in launchpad for these. I think
two bugs would be best. One for missing not being quiet, the second
for the return statuses.

https://bugs.launchpad.net/bzr/+filebug

They should generally be easy fixes if you want to get a quicker fix.

> 3)I can find no informational command like bzr missing or bzr
> conflicts that does not do anything to the branch (bzr conflicts makes
> x.BASE x.OTHER x.THIS files) except report the conflicts or missing
> revisions AND return a status indicating that these do or do not
> exist. This could easily be added as an option (e.g. --count) to avoid
> upsetting the purists who say that a return status of non-zero
> indicates an error and must not be used for anything else. But then
> bzr diff already does return just the kind of status that I hoped the
> other commands would.

Again I think a bug would be best for this.

You could also write a plugin to overcome these shortcomings, or
indeed to get the exact behaviour that you would like.

Thanks,

James