Reasons for hall monitoring
mcepl at redhat.com
Fri May 7 23:41:58 UTC 2010
Dne 7.5.2010 16:56, Przemek Klosowski napsal(a):
> Here's the rub, though: Kevin argues for aggressive development and
> empowering the package maintainers to push out changes, even if it
> resulted in temporary regressions. Ralf, on the other hand, reminds us
> about the need for quality control, process and stability. They can't
> both be right---but the entire project is better off for them voicing
> their opinions.
Well, I would argue that both can be right (although I believe Ralf
would strongly disagree with me on this point). The point in Kevin's
triades which I am afraid was utterly rejected (or ignored) is that the
most important relationship in whole Fedora land is the relationship
between package maintainer and users of the package she maintains.
When I came to Fedora from the Debian-world (around FC6 release) I was
absolutely in awe how better experience was maintaining packages in
Fedora than in Debian. It seemed like the combination of best of being
completely independent and maintaining your own repository (what would
be now called PPA; I haven't heard the term then yet) and having support
and community of fellow maintainers all in one package. It was
refreshing to see how problems were just solved on the spot without need
to apply for permission and a lot of bureaucracy. The result was
incredibly rapid development (I remember I was using as an advertisement
slogan “Fedora? Next release of your distribution today!”).
This vision in my opinion requires freedom for packagers of individual
packages to have quite wide allowance in setting their own policies
concerning updates and bug fixing. If Kevin prefers to have packages on
all distros synchronized and (maybe, I don't know, I don't use KDE)
broken much more often than Gnome-folks, it is in my opinion mostly
between KDE team and KDE users. Also, if they don't think they can
manage much more than pushing all non-packaging bugs upstream, I am not
the one who would preach to them they should do better (especially
without providing manpower to do so). OTOH, if Ralph doesn't won't to
push almost any bug upstream and he wants to make sure that all Fedora
bugs are fixed asap, great for his users, they will certainly love him,
but I am not sure it should be fixed as a rule for everybody.
This kind of "shared PPA" vision doesn't exclude some kind of stricter
requirements on critical-path packages ... if glibc or kernel is broken,
than basically everybody is affected and these components should be
fixed fast to allow others to work. But I would think that this critical
path should be really short ... glibc, kernel, udev & co. and it should
end somewhere at xorg-x11-* packages, but not much more. Certainly the
critical packages shouldn't cover by far “all what normal user needs for
his work” (including OpenOffice.org and Firefox), “whatever is on
LiveCD”, or “in the end everything” (all three ideas I have actually heard).
Of course, this kind of development process doesn’t produce distro
stable enough I could put it on my company’s server (or my mom’s
notebook), but it could be an ideal distro for developers or
contributors of any kind (with reference to
by contributor I mean anybody who provides any bits to the distro ...
packagers, artists, translators, tech writers, QA folks, yes, even
admins; I would say that somebody just handing the DVD to a friend is
doing good job in promoting the distro, but he isn't contributor in my
meaning of the word).
It seems to me that the current fashion is going sharply against this
vision of "shared PPA". There is more and more policies, permissions,
preliminary testing, etc. Suddenly packaging for Fedora is less and less
fun and more and more chore, which I don't want to dwell in. I have left
most of packaging last year (for various personal and non-personal
reasons) but I don't feel much urge to return to packaging anymore. Or
in other words, if I need some package, I can package it myself, but I
tend to keep all packages in my personal repository, and not bother with
all permissions, reviews, guildelines, policies (which is not to say
that there are no guidelines which could be helpful). I don't need that
much bandwidth and storage for my personal needs any (which is the key
resource current establishment controls). I wonder how many packagers
(or former packagers) feels the same.
More and more I was writing this email, more and more I tend to agree
with somebody today, who wrote that they key problem of the Fedora
community is unclear vision about its purpose. I agree completely. I
believe, that in the root of many of our problems lies in our
unwillingness to say that we are not end-user-oriented distribution, but
the contributors-oriented one.
There are many who pretends that it is possible to be both CentOS (or
Ubuntu) and Fedora-as-we-knew-it at the same time, at that additional
bureaucracy has no costs for development. I think it does, and that the
current trend doesn't lead to excellency in our core mission. We may end
up being the second (or even first) Ubuntu in terms of number of users.
But I really don't think the size matters in the question of how much
value we bring to the whole Linux community at all. By deserting the
position of developer-oriented distro I think whole Linux community
looses one excellent part, and in the end we may be just another
end-users oriented Ubuntu (maybe slightly better) ... who cares?
Less is more or less more.
-- Y_Plentyn on #LinuxGER
(from fortunes -- I cannot resist :-)
More information about the devel