[Nix-dev] nix channels and nix expressions

[Tony White]

2009/9/25 Tony White <tonywhite100 at googlemail.com>:
> 2009/9/25 Lluís Batlle <viriketo at gmail.com>:
>> 2009/9/25 Michael Raskin <7c6f434c at mail.ru>:
>>> -----BEGIN PGP SIGNED MESSAGE-----
>>> Hash: SHA1
>>>
>>> Lluís Batlle wrote:
>>>> The current stable releases, afaik, have only been snapshots taken of
>>>> nixpkgs before huge steps forward, and they have not been maintained
>>>> appart, so they are as stable as the trunk was at that moment. I don't
>>>> think those moments were of extraordinary stability of trunk.
>>>
>>> I think that stable releases mark the revisions you should _never_ use.
>>> It is unfortunate, but there is some reason for that: Nix-related
>>> projects are stable between inflows of fresh ideas but then there is no
>>> milestone to declare release; when there is a milestone, there is a lot
>>> of small changes before and after, so you get what you may want to
>>
>> I think that the 'stable' branch should not be declared stable as long
>> as it is not maintained apart from the main development, keeping some
>> kind of 'static-ness' to be defined. Maybe not update the pkgs to
>> newer versions? I don't know if it would be of any advantage.
>>
>> What would be the advantage of a trunk-snapshot (what we now call
>> "stable") other than directly the trunk?
>> _______________________________________________
>> nix-dev mailing list
>> nix-dev at cs.uu.nl
>> https://mail.cs.uu.nl/mailman/listinfo/nix-dev
>>
>
> Hi all,
> Maybe the build farm could do a test upgrade between the old and new
> snapshots before release, to ensure backwards compatibility, so that
> people can upgrade from snapshot to snapshot if they choose to follow
> stable?
> Then if it if it breaks before releasing the new snapshot.
>
> Does nix have the ability of prioritising itself to be downloaded,
> installed and the new version switched to before performing the rest
> of a large expression transaction which involves updating other
> packages; such as a system update?
> If yes, then that would make breaking snapshot upgrades difficult if
> they were tested before release.
>
> Thanks,
> Tony
>

Hi,
I mean ;
Then fix it if it breaks before releasing the new snapshot.
(I have the flu, sorry.)

Thanks,
Tony