duplicity-team team mailing list archive

[Kenneth Loafman <kenneth@xxxxxxxxxxx>]

On Fri, Mar 30, 2012 at 1:26 PM, Michael Terry <mike@xxxxxxxxxxx> wrote:

> Hello!  Just a note to say that apparently Ubuntu wants to try to get
> Python 2 off the Desktop CD for Ubuntu 12.10.  Which, if accomplished,
> would mean porting duplicity to Python 3.
>
> I talked briefly to Canonical's resident Python geek, Barry Warsaw,
> about it.  He recommends using the same code base (i.e. no py2to3 or
> helper libraries like python-six) and only supporting 2.6+ and 3.2+.
> One can do this by using a few 'future' imports and careful coding
> choices [1, 2].
>

Ubuntu is being overly aggressive.  At least they could put both Python 2.7
and 3.2 in the same release and give better notice to the upstream
developers.  I'm concerned that a single code base running on both 2 and 3
will be unstable and difficult to maintain.

This would mean bumping the supported RHEL version from 5 to 6.  I
> know we've talked vaguely about when that would be acceptable to do
> [3].  I'm not sure where we fell on that exactly.  But I suppose if
> Ubuntu is serious about this, they'd be willing to run with a ported
> version ahead of upstream for a while as long as that's the direction
> ya'll are going too.
>

There is a need to continue supporting RHEL 5, I think.  We should be able
to do so if we work the issues correctly. If not, then I guess it's EOL for
RHEL 5 support from duplicity.  Ten years of support for a version is way
to long in this fast paced industry.

How does that plan sound?
>

Sounds good to me.

I have the barest beginnings of such a port [4], but it doesn't really
> make any interesting changes yet.  But it will likely be where future
> work happens if anyone is interested in following along.  Nor will I
> be working on this in the near future; this is just a statement of
> intent.
>

Thanks for the start.  It's not where I would have gone voluntarily, but if
we need to do this to stay in Ubuntu, then so be it.

As to a separate port, I really would hate to support two versions.  That's
probably the reason 0.7.x has died.

Why won't you be working on this in the near future?

...Ken