bzr status

[Wichmann, Mats D]

Matthew D. Fuller wrote:
> On Tue, Nov 11, 2008 at 08:56:59AM +0000 I heard the voice of
> Russel Winder, and lo! it spake thus:
>>
>> Bazaar's insistence of showing paths from the branch root is an
>> annoyance.
>
> To provide a contrary perspective: I prefer it that way, because it
> means all the information I need to know what files are actually
> affected is all in one place [the status output], rather than having
> to be mentally combined from two [status and `pwd`], in what's likely
> to be a very lossy process (more than 2 or 3 ..'s very quickly gets
> very easy to make mistakes with).

I don't see that this is any worse really; if I'm far down the
directory tree of a branch, and I see changed files in a quite
different path, I still need to make a mental combination - with
the path-from-root of the branch.  Indeed, with relative paths,
I have something I can directly cut-and-paste into something.

Here's an example of why the path-from-root thing drives me batty:

I've got a largeish project with, among other things, some module
subdirectories, which are further subdivided by an architecture
target.  Some of the files (sgml files, these are) are hand-written,
some are script-generated by pulling from a database, but since
consumers of this shouldn't have to have the database set up to
use the project, all are committed. So this is not an unusual workflow:

... make some changes in the database ...
$ cd module-foo/generic/lib-foo
$ make gensrc
$ bzr status                 <==  ah there are new generated files
$ bzr add `bzr unknowns .`   <== nope, all the paths are wrong

Now I have to either type the filenames to add by hand, or
save the bzr unknowns output to a file and edit out the
path-from-root part before feeding to bzr add.  Not hard,
but quite irritating, when I could have let the tools do
the work for me.