Re: native NMUs
On (14/08/07 17:44), Joey Hess wrote:
> Many native packages are not Debian-specifc software, but instead
> debian-originated software (examples: dpkg, apt). Other unrelated distros
> might choose to use native Debian software. Just because it's
> debian-originated software, doesn't mean that the project as a whole is its
> "upsteam" maintainer though. Its Debian maintainer is the only upstream
> maintainer. So shouldn't he be the only one who releases new tarballs of the
> software?
>
> We could do away with the concept of NMUs of native software, and do away
> with this uncertanty, ambiguity, bugginess, etc. Simply say that when a
> NMU of a native package is done, the package stops being native in
> Debian (until the next maintainer upload, presumably). So the NMU adds
> "-0.1" to the version number, like any NMU. The .orig.tar.gz from the
> maintainer's last release is kept in the archive, with a .diff.gz added
> that conveniently contains only the NMU's changes. Since it's not a native
> package, it will be correct for the changelog to be installed as
> changelog.Debian.gz, and correct for the version number to contain a dash.
This seems to make sense to me.
However you say
The .orig.tar.gz from the maintainer's last release is kept in the archive
but for a native package there is no .orig.tar.gz, there is a .tar.gz. I
guess there are some tools that would be confused by this, in that they
wont know to unpack a source package that contains only a .diff.gz
against a .tar.gz of the appropriate version number.
Also, I guess you are opening up the possibility of creating a source
package containing both a .tar.gz and .diff.gz, which will also confuse
some tools I expect.
Thanks,
James
--
James Westby -- GPG Key ID: B577FE13 -- http://jameswestby.net/
seccure key - (3+)k7|M*edCX/.A:n*N!>|&7U.L#9E)Tu)T0>AM - secp256r1/nistp256
Reply to: