Re: Is using experimental distribution for shelter during freeze useful?

2018-11-29 Thread Paul Gevers
Hi,

On 27-11-18 12:38, Hideki Yamane wrote:
>  Well, we use experimental as "shelter" during freeze, but it's not good
>  in my point of view.
> 
>  - During freeze, it is just ignored by most of the users since they
>wouldn't know there's a newer package in there (and they also afraid
>because it's in "experimental" ;). It means "not tested" if they were
>in Debian repository for a long time period
>  - Re-uploading to unstable is just boring, and no values are added by it
>  - unstable users wants new valued packages constantly. After release,
>"package flood" to unstable is not good.
> 
>  So, I guess putting fixed packages into "testing-proposed-updates" and
>  to continue to upload packages to unstable during freeze period is better.
> 
>  Pros)
>  - unstable distribution stays newest
>  - No "unintended" changes will be introduced into testing during freeze
> 
>  Cons)
>  - Maybe you should do cherry-picking changes from unstable to
>testing-proposed-updates, not just ask "unblock" to Release Managers. 
>  - Harder to get users for test with testing-proposed-updates repository
> 
>  Your thoughts?

I am not a member of the release team, but I am involved in recent
changes to some of the tools used for migration from unstable to
testing. Since the previous release, those tools and the QA they rely on
have moved in the opposite direction of what you want and have made the
request stronger to not upload disruptive changes to unstable during the
freeze. Also see the release team's statement on this topic [1]: "It is
usually better to revert the changes in unstable and upload the fix
there.". If you want to make this an option for bullseye (I believe it
is too late to properly implement and test this all now), I propose to
help improve the QA and workflow via testing-proposed-updates, as I have
the impression that the people currently involved in the QA and workflow
(including me) are not interested to work towards your solution themselves.

So to answer your question: yes, using experimental during the freeze
for uploads not intended to be included in the next stable release is
useful.

Paul

[1] https://release.debian.org/buster/freeze_policy.html (under "Changes
which can be considered")



signature.asc
Description: OpenPGP digital signature


Re: Is using experimental distribution for shelter during freeze useful?

2018-11-27 Thread Thomas Goirand
On 11/27/18 12:38 PM, Hideki Yamane wrote:
> Hi,
> 
>  Well, we use experimental as "shelter" during freeze, but it's not good
>  in my point of view.
> 
>  - During freeze, it is just ignored by most of the users since they
>wouldn't know there's a newer package in there (and they also afraid
>because it's in "experimental" ;). It means "not tested" if they were
>in Debian repository for a long time period
>  - Re-uploading to unstable is just boring, and no values are added by it
>  - unstable users wants new valued packages constantly. After release,
>"package flood" to unstable is not good.
> 
>  So, I guess putting fixed packages into "testing-proposed-updates" and
>  to continue to upload packages to unstable during freeze period is better.
> 
>  Pros)
>  - unstable distribution stays newest
>  - No "unintended" changes will be introduced into testing during freeze
> 
>  Cons)
>  - Maybe you should do cherry-picking changes from unstable to
>testing-proposed-updates, not just ask "unblock" to Release Managers.

The process would stay the same, just instead of uploading to unstable
during the freeze, we would upload to t-p-u.

>  - Harder to get users for test with testing-proposed-updates repository

Nothing would prevent one from uploading to both t-p-u and sid.

I very much support this proposal, but we're probably too close from the
freeze already, and this would probably also need some work on the
release team and/or FTP master side. If you want this to happen, maybe
you should get in touch with both teams directly and do the work *after*
buster is released? Anyway, they would tell...

Cheers,

Thomas Goirand (zigo)



Re: Is using experimental distribution for shelter during freeze useful?

2018-11-27 Thread Sean Whitton
Hello,

On Tue 27 Nov 2018 at 08:38PM +0900, Hideki Yamane wrote:

>  Cons)
>  - Maybe you should do cherry-picking changes from unstable to
>testing-proposed-updates, not just ask "unblock" to Release Managers.
>  - Harder to get users for test with testing-proposed-updates repository

- Additional work for the release team as the testing-proposed-updates
  mechanism is more work to approve changes, AIUI.

-- 
Sean Whitton


signature.asc
Description: PGP signature


Re: Is using experimental distribution for shelter during freeze useful?

2018-11-27 Thread Dominik George
>  Your thoughts?

sid is not a rolling release for the public, it is a development area.
Some users use it as a rolling release to get bleeding edge software,
but in fact they become a developer that way (not meaning DD).

If you think regular development prevents you from staying up to date
during the freeze, install the packages you need from experimental. You
are a developer, after all.

-nik


signature.asc
Description: PGP signature


Re: Is using experimental distribution for shelter during freeze useful?

2018-11-27 Thread Jonas Smedegaard
Quoting Hideki Yamane (2018-11-27 12:38:46)
> Hi,
> 
>  Well, we use experimental as "shelter" during freeze, but it's not good
>  in my point of view.
> 
>  - During freeze, it is just ignored by most of the users since they
>wouldn't know there's a newer package in there (and they also afraid
>because it's in "experimental" ;). It means "not tested" if they were
>in Debian repository for a long time period
>  - Re-uploading to unstable is just boring, and no values are added by it
>  - unstable users wants new valued packages constantly. After release,
>"package flood" to unstable is not good.
> 
>  So, I guess putting fixed packages into "testing-proposed-updates" and
>  to continue to upload packages to unstable during freeze period is better.
> 
>  Pros)
>  - unstable distribution stays newest
>  - No "unintended" changes will be introduced into testing during freeze
> 
>  Cons)
>  - Maybe you should do cherry-picking changes from unstable to
>testing-proposed-updates, not just ask "unblock" to Release Managers. 
>  - Harder to get users for test with testing-proposed-updates repository
> 
>  Your thoughts?

Let's not make freeze more comfortable.

Let's make freeze more efficient - and uncomfortable to ignore.


 - Jonas

-- 
 * Jonas Smedegaard - idealist & Internet-arkitekt
 * Tlf.: +45 40843136  Website: http://dr.jones.dk/

 [x] quote me freely  [ ] ask before reusing  [ ] keep private


signature.asc
Description: signature