x86_64 FC2 repo not up yet?

[Joe Harrington]

[In rereading the text below, it's rather long, for which I apologize.
The main point is that there's a mismatch between impression and
reality on the Legacy web site, and I don't think it's in anyone's
long-term interest to perpetuate it.  It certainly isn't in the
user's.  I think part of why it's so long is that I'm trying to be
kind and make a strong point at the same time.  Please understand that
I do hold the project and its volunteers in high regard, and that I'm
not looking for huge changes.  But, to get the changes right may
require folks to develop an outsider's point of view on the project,
which is what I'm trying to give.]

> So, it isn't that we are not supporting it, it is that until recently there
> was no need to support it, and now that there is, we've not yet set up any
> infrastructor for it, and we don't yet have enough people wanting it and
> willing to support it.

Well, we're getting mixed up in semantics.  All I said was that you do
not support x86_64, not that you won't ever.  It is true that you do
not currently support it, regardless of the intentions, and you also
reserve the right not to support it ever if the resources are not
there.  My offline conversation with Jesse Keating earlier today
indicated some doubt as to whether the resources were there.  And,
we're not talking about a trivial interval of non-support, given that
a security bug was reported March 28 (CAN-2005-0468 and CAN-2005-0469,
thanks to David Eisner for pointing this one out to me in offline
email).

The reason I'm pointing out what might otherwise be a nitpick is that,
regardless of the lack of a formal financial/contractual commitment to
users, a lot of people look to this project to see whether they can
use Fedora Core in real-world environments.  Many people ask me, is
Fedora stable if you pick it up two releases behind and run all the
way through Legacy?  You are looked to as an avenue to a stable
Fedora, and strong statements that are backed by plans rather than
reality can lead people to make bad decisions.

> Well, it is the case, but only if we have sufficient community 
> interest/support/involvement.  And, we may not be able to support
> everything on day one, sometimes there may be a delay in getting
> support for something.

You're making my point beautifully.  Since things are not necessarily
supported right off, and not everything is going to be supported, you
should make this clear in plain language (think Warren Buffett) right
on the front page, and give people something like a chart or a list or
a per-release status page so that they can see 1) what is currently
supported, 2) what you're actively working on getting supported, and
3) by extrapolation (and at their own risk), what they might expect
for a future distro.  In my case, had there been prior releases with
x86_64 versions, and had I seen Legacy not support them, I would have
decided to go with something other than Fedora for my cluster, and it
would have been a good decision: My cluster has been subject to a
known security hole for almost a month now, due to the lack of
updates.  That scares me, and while I see good intention here I think
it will be at least a few more weeks more, and possibly much longer,
before support arrives.

> And secondly, the implication you cite is just the
> implication we want, though at times we may drop the ball and be
> unprepared or get to things in a less than timely fashion.

Careful here, you're saying you want people to believe something that
you admit may not always be true.  I'm not sure where that sits in
your moral framework.  It doesn't sit, in mine.  This isn't some toy
you're dealing with, it's something depended on by thousands of
people, and real damage measured in real dollars and real hours occurs
when someone gets broken into.  People decide whether to depend on
Fedora based on your statements and those on the Fedora site.  To be
credible in the long term, you need to make up for your lack of formal
liability by being brutally honest and transparent at all times.

> You only need read the main web page where it says the transition of
> FC is underway but not yet complete.

> No, the statement that we've not yet transitioned FC2 to FLP clearly
> states that not everything is maintained yet for FC2.  And many places
> we discuss that support for anything requires community support/involvment/etc.

You misunderstand me here.  I'm talking about someone who has a new
box and is deciding what to put on it.  This person presumably starts
with a current version of FC, and checks Legacy to see whether they
will in the future have support, even after Fedora EOLs it.  For me
that was last July, and of course there was nothing about the current
transition.  I was looking for statements relevant to whether x86_64
would be supported.  I don't recall whether the x86_64 item was on the
FAQ then, but I do recall coming away satisfied that your commitment
to all the releases was high enough that I'd see a relatively seamless
transition.  That is the impression you gave, quite effectively.

Someone doing that today, who might start with FC3 or wait for FC4,
would have little to indicate that taking their release into Legacy
would be anything other than a change of repo in yum.conf.  It would
be more honest and transparent if you wrote (and kept up after the
current transition was complete) that the most recent conversion to
Legacy took a month (or whatever it ends up being), in which time
users of those systems had an open security vulnerability, and that
you hope, but do not guarrantee, that the next transition will be
quicker (and sure, give a plug for more volunteers).

I'm not trying to rain on your parade here, though I think my words
will have that effect.  I'm pointing out that there is a big gap
between the overall impression given about Legacy support and the
reality.  I'm also not laying blame or even saying that you have not
done the best you can with the resources you have.  I'm saying that
the users can only care about the end result, and you serve them,
Fedora, and yourselves best if you are quite clear and open about what
you can in practice actually provide.  It's not the place for
marketing-like glossovers of weaknesses.  It is better to make an
honest impression than a good one.

> Sounds about right.  Might you be interested in participating?  I don't
> remember seeing a "Self Introduction" or other notice from you... ;)

Well, it will be intro and conclusion.  I joined the list today after
both Jesse Keating and Matthew Miller suggested it, just to ask the
two questions I have asked.  I'm currently coordinating part of the
development of scipy (the "let's write some real docs and make a real
web site and make some binary packages now that the software is coming
together" part), and I'm afraid that other than filing bugs, I can
only do volunteer time on one project.  I'm also searching for a
faculty job in astronomy (I study extrasolar planets), and I'm the dad
of a toddler.  Given the latter, there's also a lot of "I used to"s I
could add, but won't, mainly because life-with-a-life is a fading
memory.  For background, I cut my computational teeth as a systems
developer for MIT Project Athena (think X and Kerberos, though I
didn't work on those projects except as a "friendly tester") in
1986-1988, turning the first workstations into things more useful than
a Vax 750.

--jh--