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

Re: [PROPOSAL] don't upload to "stable unstable" (was Re: BAHH. Retraction.)



[cc me if you have any comments, I don't read -policy... especially not
over a modem for the holidays...]

On Wed, Dec 27, 2000 at 03:46:48PM -0500, Ben Collins wrote:
> PROPOSAL:
> 
> With the integration of the "testing" distribution, we will not (in my
> understanding) ever upload to frozen. Our uploads will generally go to
> unstable, and propogate to "testing" based on some rules (bugs, time,
> etc..).
> 
> Currently, it is still possible to build a package and upload it to
> "stable unstable". A lot of discussion is bringing to light just how bad
> this is. It's something porters have had to deal with for quite some
> time. What I am proposing is that uploads to stable can never have a
> newer version than the same package in unstable. Basically, do not allow
> uploads to go to "stable" and "unstable" using the same binaries. So if
> we have this setup:
> 
> stable:   foo_1.1-1
> unstable: foo_1.1-2
> 
> If a new version is required for "stable" (because of a agrave bug,
> secrity fix or what have you), then it should get an NMU type version
> increase, such as "foo_1.1-1.1". If The new upload requires a new
> upstream version, then the stable version should get "1.2-0.1", while
> the unstable version is simultaneously (or previously) updated to
> "1.2-1".
> 
> I'm interested in hearing corner cases, so that they can also be covered
> in a sort of multi-upload-howto. IMO, stable updates are not frequent
> enough that this would be really hard.

The only legitimate use for stable unstable uploads right now is for
packages which:
  - Have not changed since stable was released
  - Have some bug important enough to merit a fix in a future revision
    of stable.

(or have absolutely non-backportable fixes)

I would say that asking anyone who is going through the process of
uploading to stable to do a second upload just for it is a very small
price for the clarity this would gain us.

Dan

/--------------------------------\  /--------------------------------\
|       Daniel Jacobowitz        |__|        SCS Class of 2002       |
|   Debian GNU/Linux Developer    __    Carnegie Mellon University   |
|         dan@debian.org         |  |       dmj+@andrew.cmu.edu      |
\--------------------------------/  \--------------------------------/



Reply to: