dulwich-users team mailing list archive

[Dave Borowitz <dborowitz@xxxxxxxxxx>]

On Mon, Dec 27, 2010 at 19:08, <jelmer@xxxxxxxxx> wrote:

> On Mon, Dec 27, 2010 at 05:44:01PM -0800, Dave Borowitz wrote:
> > Looks like only 3 of these got picked. Anything I need to change, or have
> > you just not gotten to the rest?
> I tried to land the 4 that were obviously correct to me. I was meaning to
> reply
> about the other ones but hadn't gotten round to that yet.
>
> > On Tue, Dec 21, 2010 at 14:50, <dborowitz@xxxxxxxxxx> wrote:
> > > I've been meaning to help knock some bugs off the tracker for a while,
> > > so here are some easy fixes.
> > >
> > > There are a number of other outstanding bugs I'm still working on, but
> > > which will require significantly more code.
> > >
> > > ebc736f test_repository: Add setUp/tearDown to CreateRepositoryTests.
> Is this preparing for more tests in that TestCase that hit disk, or perhaps
> just simplifying the test method?


It's for the benefit of the next patch. I would have had to add try/finally
blocks to all 4 of the new test methods, so I simplified it ahead of time.

BTW, it's clear from our other discussion that this patch isn't correct in
that setUp/tearDown need to call the super methods as well.


>  > > 0c56328 repo: Allow recursive mkdir in Repo.init(_bare).
> I wonder if this is really the right layer for this kind of thing. Wouldn't
> it make
> more sense to put that sort of thing in the UI layer?


I'm not sure what you mean by the UI layer? IMO if we allow making any new
directories with mkdir=True, then we should allow making new directories
recursively. If you think making directories at all shouldn't be within the
scope of Repo.init, then we should remove the mkdir arg entirely, no?


> > > 8cc2aee client: Standardize on single quotes.
> I'd like to land this patch, but I also think we should decide on a
> standardized way to deal with quotes. Should we always just use single
> quotes and
> only use """-style for docstrings? IIRC PEP8 doesn't say anything about
> this,
> so we should probably update HACKING with whatever we decide.
>
> > > c6c8547 test_client: 80 cols.
> This one had too many conflicts as I didn't land some of the patches
> preceding it.
>

Fair enough :)


> Cheers,
>
> Jelmer
>
> > > 51134ed client: Rewrite URI parsing in get_transport_and_path.
> > > 82b38c4 tutorial: Clarify setting of HEAD.
> > > 21e6b47 repo: Raise RefFormatException instead of KeyError on bad refs.
> > >
> > >  docs/tutorial/2-object-store.txt |   22 ++++++++--
> > >  dulwich/client.py                |   65 +++++++++++++++++-----------
> > >  dulwich/errors.py                |    4 ++
> > >  dulwich/repo.py                  |   14 ++++--
> > >  dulwich/tests/test_client.py     |   86
> > > +++++++++++++++++++++++++++++++++----
> > >  dulwich/tests/test_repository.py |   72
> ++++++++++++++++++++-----------
> > >  6 files changed, 193 insertions(+), 70 deletions(-)
> > >
>
> --
>