Re: Documentation on handling of orig.tar.gz files for Developer's Reference or for Debian Policy
On Mon, 1 Nov 2004 14:04:44 -0800, Chris Waters <xtifr@debian.org> said:
> On Mon, Nov 01, 2004 at 01:47:02PM -0600, Manoj Srivastava wrote:
>> How about this rule of thumb: If you get stuff from the primary NON
>> DEBIAN distribution site, that is what you call upstream. What
>> someone unconnected to debian, not using debian archives, downloads
>> is what we also offer as upstream orig.tar.gz files.
> I think it's more important that our users *know what they're
> getting* than that we try to enforce some sort of arbitrary
> distinction between "Debian" and "upstream".
And so they should. Every source packages from debian should
have that information. (And I mean source package formally -- .dsc
and friends).
> Clarity is why I chose "107.0pre108" as a version number. I don't
That is another red herring, though it is good you selected a
clear version number.
> see how it's that much different from our various cvs-snapshot
> packages, except that in my case, upstream wasn't using any sort of
> version control at the time I made the package - they just had a
> loose collection of patches and replacement files available on their
> website.
Umm, CVS snapshot packages have a clear version number as
well -- which again is good -- but is irrelevant to mangling upstream
source tar balls.
>> Pristine upstream means pristine upstream. Either get your notes
>> added to upstream website, or put them in the diffs.
> We don't require "pristine upstream". For example, we remov
Did I say we required them? We do, however, recommend that we
ship pristine upstreams as far as possible.
e
> non-DFSG compliant portions. Many licenses require that changes be
> documented. So if we modify the upstream source to remove the
> non-DFSG portions, and _don't document that_ (because of a new
> policy rule that forbids any debian-authored portions of _orig
> tarballs), then we may be violating licenses.
Bravo, for belabouring the obvious.
>> Do not prevaricate to our suers by pretending that some material is
>> the same as they can get upstream, when it is not.
> I don't think I am - I think it's quite clear that 107.0pre108 is
> quite different from 107.
But you are stuffing material in there that us not in the
upstream sources. If merely versioning makes itr clear to users that
things are different, why do you need to mangle the upstream by
adding material in there?
>> > Anyway, I was upstream project leader for most of the
>> > last year, up until about a week ago, when I stepped down in
>> > favor of someone more enthusiastic. But I'm still an upstream
>> > developer.
>> That is quite irrelevant.
> Actually, I agree. I think the fact that I can solve "the problem"
> by sticking the tarball I made on the upstream website at any moment
> I choose is, or should be, irrelevant. I think the tarball I
> created should be acceptable in any case. I think it's quite clear
> what I created, and I don't think there's any intent to confuse our
> users, and I think that should be sufficient.
Just because you can fix the situation by retroactively
modifying upstream does not alter the fact that you have chosen to
add material to a file you call orig.tar.gz , which, I think,
violates at least the principle of least surprise.
manoj
--
The test of intelligent tinkering is to save all the parts. Aldo
Leopold
Manoj Srivastava <srivasta@debian.org> <http://www.debian.org/%7Esrivasta/>
1024D/BF24424C print 4966 F272 D093 B493 410B 924B 21BA DABB BF24 424C
Reply to: