[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: DEP5: reasons for not pushing Bazaar changes to Subversion

On to, 2010-12-23 at 09:55 +0100, Stefano Zacchiroli wrote:
> So, let's get back to the basic principles at stake. The point hardly is
> Bzr vs SVN; rather it seems to be whether the working draft of a DEP
> should be constantly updated and trivially accessible, for instance on
> the web at the canonical URL listed at the beginning of the DEP text.
> 1) The argument in favor of that approach is: it diminish to the very
>    minimum the barrier for all people interesting in contributing.
> 2) The argument against that, which has been mentioned earlier on in
>    DEP-5 discussions, is: it encouraged proliferation of adoptions of
>    working drafts, creating an incompatibility mess.
> In general, I find that the benefits of (1) largely outweigh those of
> (2). The status of a DEP is clearly marked and if people would like to
> adopt a DRAFT, well, they are on their own in keeping up with the
> evolutions of the format later on.

I entirely agree with this. In the general case DEP drafts should be
public, even aggressively so.

> In the specific case of DEP-5 however, I understand the sensibility of
> the drivers about (2). Half-baked versions of the format, not really
> controlled by anyone, have been floating around for a long while. Even
> more so, at some point there has been a sort of encouragement in
> adopting ever-evolving version of the format by pointing to a specific
> revision of it. So, while I believe that advantages of (1) outweighs
> those of (2), even in the case of DEP-5, I can't really blame the
> drivers for having chosen a more "conservative" approach.

I agree with this as well. Further, quite a number of people were
against the entire idea of a machine-readable debian/copyright files,
and it seems to me that was due to the way the spec was being developed
over the years. The fact that it took years was, itself, also a problem.
This should only take a few weeks, at most. (It's taken five months
already since I became a driver, for which I apologize. I should have
made things go faster.) At least the most vocal complaints about the
DEP5 concept and format seem to have died away recently. It's even been
a while since anyone offered DEP5 as an example of why the entire
concept of DEPs was unworkable.

This sensitivity within the Debian development community to the issue is
the other reason I don't want to have lots of versions of *this* spec in
actual use. It's not just the practical effort that's going to be
required to update from umpteen different formats to the final one, it's
also that I want to minimize any further the aggravation from this.

Now I have failed to do that, in another way.

I also agree with Charles that we're about ready to finish the DRAFT
stage, and can stay with status quo until then.

Blog/wiki/website hosting with ikiwiki (free for free software):

Reply to: