Le lundi 25 octobre 2004 à 12:08 -0400, Paul Iadonisi a écrit :
On Mon, 2004-10-25 at 10:06, Matias Féliciano wrote:
[snip]
> Please... I not chose to *ignore* those warnings.
> You are ignoring that using a test release is the _best_ way to test it
> and be fully happy with the finale release when it's out. The point is
> not to be happy with a test release but with the finale release.
>
> The propose of test release can be see as :
> - find stupid boys (like me), or girls, which use test release and file
> bug report.
> - provide some support (bugfix, mailing, etc) to not make it a total
> disaster.
Well, I suppose that's an okay way of looking at it, but methinks you
need to make up your mind. Are you ignoring the warnings not to use
this on a 'critical' system, or is your desktop that you have it
installed on *not* a critical system.
My desktop is a critical system for _me_. It's up to me to decide what
to install on it (and thank to Red Hat to inform me about what this
imply).
btw, I have not replace FC2 by FC3T2. First I install FC3T2 along my
rock solid FC2. I test FC3T2 and when I have the feeling that FC3T? is
solid enough, it become my default desktop and next I remove FC2.
To use or not FC3T? for a "critical-mission" (like my desktop) is up to
me.
off-topic: my desktop is critical to me because I have only one PC at
home and use it for Internet, mail, watching/recording TV/DVD, desktop
stuff...
Can't have it both ways.
I was about to say that this has strayed far off topic, but adopting
good testing practices seems pretty on topic for fedora-test-list.
It is true that the best way to test it is to use it. I think that's
pretty much a given. And it's fine, and actually encouraged for testers
to use it on a desktop they use on a regular basis.
But rely on it to be secure or stable
I _never_ said that.
and developers will probably point and laugh at
you. And justifiably so. Of course they'll want to hear about security
problems and other bugs. After all, that's the whole point. But if a
test release eats your critical data for which you have no backups,
Beta/test or not, you _always_ need good backup. I have *2* backups
(nightly build). I archive all rawhide changes if I need to revert to an
old package. If some thing does wrong with the last rawhide, I can boot
with my previous backup. If it doesn't work, I can use my previous
previous backup.
or
messes up your desktop so bad that your late on a project at work, no
one is going to have any sympathy for you. Put it on a server serving
hundreds or thousands of users who are not willing testers and
developers will call you insane.
I apologize if I seem a bit antagonistic about this. It's just that I
see it as pretty important that people understand what testing means:
I understand what testing is. I use beta/test release since RH8.0. I
never complaint about it eat data.
that your data may get eaten alive and no one will be able to save
you
from that. That's at least partially what's meant when Red Hat says not
to use it on critical systems.
And it's also important to understand what the Fedora Project is and
how it differs from RHEL.
RHEL *absolutely* *will* receive more attention from Red Hat than
Fedora Core.
What are we talking ?
We are talking about beta/test.
I feel Red Hat put the same level of attention to Fedora test than they
put in RHEL beta. Perhaps a "different" attention.
Please, a Red Hat employer can confirm this ?
RHEL is the money maker.
Yes. And Fedora help improving RHEL (don't forget this). I think Red Hat
do a very good decision with the "couple" Fedora/RHEL.
But this decision does not imply that Red Hat should have two class of
testers :
- First class testers : RHEL (rpm signed, etc)
- second class testers : Fedora, don't care about this "class" of testers.
Let me try to summarise (in my bad English).
It's up to me to decide to use a beta for a mission critical computer.
Right now, for my personal computer, I feel the risk is pay back because
this improve (I hope:-)) Fedora (and RHEL in a long run). And also
because I like enjoying with the latest technology.
By not signing their rpm in rawhide, Red Hat "force" me to take risk
(fake rpm, ...) for _nothing_. I don't want to take these risks.
Fedora Core is the proving grounds. That's the bottom line. At
least,
that's how I see it.