On Fri, Jun 22, 2012 at 11:01:56PM +0200, Martin Zobel-Helas wrote:
> The current practice for debian.net entries is that they are directly
> entered in the debian.net zone as 3rd-level records. I am seeking
> comments on a proposal to alter this practice.
Thanks for working on the idea!
Here are a few comments of mine:
- I like very much the idea of thinking at *.debian.net entries as
services in "incubation". It is way clearer than our current notion of
"official" vs "non-official" (hosting), which is something that people
who have no idea what "DSA" is simply cannot understand. Additionally,
it bring purpose to debian.net entries as something temporary that
should at least ideally strive to move to debian.org. Not because our
DSAs are control freak (they're not), but rather because debian.org
hosting gives more guarantees to the Debian Project of long term
maintenance, usually because it reduces the bus factor risk.
Making this even clearer with a *.incubator.debian.org namespace might
be a good idea. (Modulo some transition time, doing so will eventually
replace *.debian.net, if I got that right.)
- I've already discussed in a related thread of a few months ago how I
think the current distinction between debian.net and debian.org should
be documented, incidentally resolving other visibility problems of
those services. Not that the dnsZoneEntry LDAP entry is publicly
available, we should have an automated generated index of debian.net
services, with pointers to the responsible DD. I think it'd be a good
idea to have such index live at http://www.debian.net together with an
explanation of the debian.net/.org distinction. I don't think *this
part* of the confusion is enough to justify changes of the current
scheme (but see below for another possible reason).
Stuart Prescott (Cc:-ed) has already drafted an implementation of the
index generation and I've encouraged him to submit it as a wishlist
bug report + patch to the -www team a few weeks ago. Stuart: any news
on that front? (I can't check if the bug report is on right now due to
shaky mobile phone connection.)
- Regarding *.$user.debian.net vs *.debian.net, I think both have value.
The former might be useful for DDs that want to offer some kind of
service, but explicitly want to avoid "polluting" a namespace that
makes services *look* somehow blessed by the Debian Project. That is
an attitude we should cherish and support. If we go for the
incubation principle, it will also be useful to explicitly mark which
services are /not/ in incubation, but only personal hacks that will be
perceived as less "stable".
The latter (*.debian.net) is important for the opposite reason:
explicitly communicate that a service is somewhat tied to the Debian
Project, even though not officially hosted on its infrastructure
(yet). Unfortunately, your proposal is silent on who/how will decide
when a *.debian.net entries is acceptable or not. It's hard to comment
on that :-), but I think it's crucial to have a proposal about it
before deciding on the more general scheme you're proposing.
Generally speaking, every time we add an approval step I start to fear
bottlenecks and the creation of new mighty powers; avoiding that is
one of the key advantages of the current scheme. If the main problem
is "squatting", then I see two possible solutions:
1) be liberal by default, but empower someone to decide that a name is
not acceptable. I think DSA would be a reasonable choice, as you
already decide on *.debian.org, but I suspect DSA would not want to
have this veto power (choice which I respect)
2) find some clear cut rule. One I've proposed in the past is that for
any *.debian.org entry, the corresponding *.debian.net should not
exist (or point to the debian.org ones, depending on the protocol).
This one will still give some sort of veto power to DSA, but it
will come with the factual justification of an existing homonymous
service
Technically, I also fear the collisions between *.$user.debian.net and
*.debian.net: the latter is open ended, as well as the user namespace
in Debian.
Cheers.
--
Stefano Zacchiroli zack@{upsilon.cc,pps.jussieu.fr,debian.org} . o .
Maître de conférences ...... http://upsilon.cc/zack ...... . . o
Debian Project Leader ....... @zack on identi.ca ....... o o o
« the first rule of tautology club is the first rule of tautology club »
Attachment:
signature.asc
Description: Digital signature