motu-release will revert libgems-ruby to the old state.
Stephan Hermann
sh at sourcecode.de
Wed Sep 3 01:30:38 BST 2008
Moins,
be careful, it could be that I'm tramping on someones foot, whoever take
this personally, damn, you deserved it ;)
On Tue, 2008-09-02 at 17:40 +0200, Soren Hansen wrote:
> On Tue, Sep 02, 2008 at 12:08:02PM +0900, Emmet Hikory wrote:
> > [...] Mathias is in agreement that MOTU Release has the authority to
> > supercede uploads that may be considered inappropriate for a release.
>
> Well, I'm not sure I am. The closest to a charter for MOTU-release I
> could find is https://edge.launchpad.net/~motu-release that says:
>
> This team takes care of approving and denying Feature Freeze
> exceptions for Universe and Multiverse.
>
> I don't see how reverting technical decisions done by fellow MOTU's
> falls under this authority.
>
> While every MOTU has the power to revert any other MOTU's changes, a)
> it's bad manners, and b) it's something entirely different if it's a
> team that invents its own authority to trumph "a lowly MOTU"'s upload.
Whatever team is there, I think they don't have more power then one
MOTU. All developers with upload powers belong to motu, despite the
fact, they have core powers.
If this is only a power question, ok, one pig is more equal then other
pigs, especially if they are being paid for being more equal pigs.
If it's a question of knowledge and experience, then I think most people
in the team should go away, until they claim >=15 years of work in
sysadmin areas and doing software development since the 80ties.
Most people don't and they act only on personal matters. They had a
touch of RoR, and they found the installation system of Gems on debian
or ubuntu horrible, "let's fix it for me, i has da powers".
Most work on Ubuntu Main/Universe or Debian is done because people are
interested. But, sometimes those work is not needed for most people (but
the change is good), or totally wrong.
This gem upload was wrong.
If Lucas would have uploaded that to Debian, I would tell him the same,
and I would have to find a sponsor to revert that upload.
> I'm happy to engage in a discussion as to whether motu-release *should*
> have this authority. In fact, I'm inclined to think that they should.
> I'm simply pointing out that they currently do *not*.
As said, the upload wasn't even discussed with upstream, which is in our
case Debian. Lucas already objected, and for me, it's enough to revert
the change.
This change had to be discussed with upstream and upstream of upstream,
because it's not a problem of Ubuntu, and not a problem of Debian, but
like ez_setup for python, a problem of ruby upstream. We need to focus
on fixing the wrong view of ruby upstream, and while this is unresolved,
we need to take care of problems. Sym/Hardlinking/copying
from /var/lib/gems/* to /usr/local/* is wrong and dangerous without
notifying the sysadmin, and even with a notify it's wrong and dangerous.
>
> > Separately, I would like to personally encourage all developers to
> > [...] and request to MOTU Release to consider the option of a longer
> > discussion period prior to superceding an upload except in cases where
> > such a delay may adversely impact some specific release milestone.
>
> I'd like to also stress this very thoroughly.
>
> I strongly disapprove of motu-release's conduct in this matter, but done
> is done, and I'm not going to revert their reversion by uploading yet
> another gems package to do something that I think is better, shinier, or
> whatever. I will just be very disappointed if motu-release again decides
> that it has authority to carry out a maneuver like this, and to make
> matters worse, doesn't leave any time at all for appeal.
TBH, if I would use openldap nowadays (which I'm not), and the package
would ask me to convert from slapd.conf to cn=config without having a
way to stop that, I would even revert that upload. But before I do that,
I would ask the TB and/or MR to review the technical decision made by
someone who has no clue.
Some things a sysadmin should decide. If we take the workload from
him/her, ok, but taking essential decisions from him/her, this is a no
go.
And packages like gem or openldap are not packages which are installed
without purpose (openldap is just an example here).
I mean, colleagues, we are talking about servers here, not about totally
cluttered desktops of people who are plastering their system with unused
packages.
I really wouldn't care if gem would be a desktop package which could
install the honey bunny skyper diaper hyper apps from
"cluttermydisk.com".
That means, I do like people who object other peoples decisions. The
question was raised on this and u-m,u-d mailinglist and many people
objected. At least two or three people were in favour of the change,
just because they had to (think of your team lead in your company who
doesn't favour you and your decision to the public...bad team lead I
would say).
I don't think that many people in those threads ever touched gem in a
commercial environment, if they did, they wouldn't even touched that
piece of software crap ever again and they would fire those people using
it.
After all, ubuntu-server is a business usecase, and imho we should
respect the business. Sysadmins know how to set symlinks or how to use
the system. So don't take them the decisions what to happen to crappy
software installed via gem.
Kind regards,
\sh
PS: be careful, it could be that I tramped on someones foot, whoever
took that personal, damn, you deserved it ;)
PPS: But really, promised, there is no personal attack attached and
never will.
More information about the Motu-council
mailing list