Storage model comparison ditz/pitz with git-issues

[Jeff Kowalczyk]

Hi Matt,

Thanks for starting up this project, I was interested in ditz, but not
using it yet. I am much more interested in a python implementation,
and seriously considering using pitz. Glad you chose YAML as the data
format.

Have you compared the storage model of ditz/pitz (YAML files with a
GUID filename) to that of git-issues [1] i.e. XML (could easily have
been YAML) using a SHA name in 2/38 character folder heirarchy?

I realize the latter is a git-ism which would impede cross-vcs (or no
vcs) use. But the idea of content (SHA) as issue file name was
interesting in contrast to GUIDs.

I was wondering how effectively ditz issue files worked in DVCS
scenarios, and whether the DVCS's were handling merges automatically
among various repositories' collections of ditz issues.

Thanks,
Jeff

[1] http://github.com/jwiegley/git-issues/tree/master

[Matthew Wilson]

Hi Jeff,

On Wed, Mar 4, 2009 at 5:04 PM, Jeff Kowalczyk <jeff.ko...@gmail.com> wrote:
>
> Hi Matt,
>
> Thanks for starting up this project, I was interested in ditz, but not
> using it yet. I am much more interested in a python implementation,
> and seriously considering using pitz. Glad you chose YAML as the data
> format.
>
> Have you compared the storage model of ditz/pitz (YAML files with a
> GUID filename) to that of git-issues [1] i.e. XML (could easily have
> been YAML) using a SHA name in 2/38 character folder heirarchy?

I don't understand the 2/38 character folder part, but I don't really
like the idea of using XML because using plain-text diff tools with
XML is no fun at all. Also, I want to support the use case where
somebody just edits stuff manually with vi.

> I realize the latter is a git-ism which would impede cross-vcs (or no
> vcs) use. But the idea of content (SHA) as issue file name was
> interesting in contrast to GUIDs.

Would the SHA change as the content of the issue changed? In git, as
far as I understand, the SHA comes from the patch, which is immutable.
But an issue is likely to have lots of revisions. So I'd rather not
use a hash for it, because as soon as I change the spelling, or update
a date, the hash is meaningless.

> I was wondering how effectively ditz issue files worked in DVCS
> scenarios, and whether the DVCS's were handling merges automatically
> among various repositories' collections of ditz issues.

I haven't had any real-world experience merging ditz conflicts, but I
figure that it would be about as difficult as manually merging in
source code.

Thanks for the feedback!

--
Matthew Wilson
ma...@tplus1.com
http://tplus1.com