On Fri, 26 Sep 2008, Dag Wieers wrote:
If I would be giving the wrong impression by indicating all is well,
how would that be better ?
It wouldn't. What I meant to express in my previous messages was that
communication is very important. You are creating the packages and
maintaining the repo and you know before any of the users if some
problems are present in the packages. You may choose to hold some
updates back, release them knowing that they break some other packages
until you have the time to rebuild those or release them knowing that
they break other packages and nothing can be done about it - it
doesn't matter what the situation is, if you know _and_ you let people
know what the status is they can plan accordingly.
You already make available the build logs, which is a great resource
for seeing how something is built (f.e. what configure options) or the
errors that appear. But build logs are not easy to monitor for such
build errors - that's where some automated tools that check the
results of the build and the state of the repo could help a lot. These
tools should run on your side and would also allow you to add some
words about the decisions you make; running it on the users' side is
not efficient and prone to errors anyway (f.e. because of mirrors).
The problem is that RPMforge does not scale very well.
Given the huge number of packages this is not surprising. Actually I'm
surprised how well it worked so far with so few people ;-)
We are a bunch of guys, there is not central infrastructure.
I think that this is the main issue: it grew without you putting too
much effort into organizing it and you are now hitting the limits of
what the current way can do. Maybe time for a change ?
But doing what we have been doing is not viable.
Because it depends on me mostly.
So we need to remove the dependency on me, and that is one of the reasons why
I don't want to change what we have now (the other reason is the time I am
spending right now).
I don't quite follow you here... what you're saying contradicts
itself.
I am not saying people have to switch to apt. But it would be one
solution. The other is helping with the rpmrepo project, or making
yum more resilient.
Making yum more resilient is already possible - I'm sure that it can
also be improved - but my point in another e-mail was that this
behaviour is not the default one in the many distributions that use
yum as their updating mechanism. As some of the RHEL/CentOS/SL
distributions will have to be maintained for some more years in
more-of-less the current form, it's not reasonable, in my view at
least, to expect all of them to change to a more resilient yum just as
it's not reasonable to expect all users to switch to apt.
As to rpmrepo, I see that there is some activity on its -devel list
now. Maybe this dialogue has shown that there is interest in the
switch to rpmrepo; if so, then I'm glad to have been part of it :-)
If nothing happens, that trust will be lost at some other point in
time. Your trust does not feed me.
Indeed, but I was hoping that trust from your repo's users gives you
an incentive to continue doing it.
I cannot promise that they will be fixed shortly. Do you prefer that
I lie ? I may be able to update flac, but who knows what problems
that may spawn.
I prefer to know about the problems when you discover them or about
decisions when you make them. For me, it's OK to say:
audacity cannot be updated right now because it depends on an
older version of wxGTK which is no longer available
because...
Users who want to continue using audacity should refrain from
running 'yum update'; those who can switch to another similar
software could remove audacity ('yum remove audacity' or 'rpm
-e audacity') then continue to be up-to-date by running 'yum
update'.
We'll announce if/when a different solution is found.
Apt users don't need to be bothered; apt is able to work
around this by...
It allows people to know what is going on, gives an immediate solution
and the information that a better solution is being looked for. Also
gives a plug for apt ;-)
I released them because they were blocking other updates after a few
days of trying to find a solution.
And it's fine to do this now and continue to do this in the future as
long as the users know about what is going on.
Thanks again for all your work !
--
Bogdan Costescu
IWR, University of Heidelberg, INF 368, D-69120 Heidelberg, Germany
Phone: +49 6221 54 8869/8240, Fax: +49 6221 54 8868/8850
E-mail: [EMAIL PROTECTED]
_______________________________________________
users mailing list
[email protected]
http://lists.rpmforge.net/mailman/listinfo/users
