Alan Cox wrote:
The _only_ reason I will admit for its use as a server is for software
development, where one is developing client or server against it,
targetting a future release of RHEL. Then, I would use it as needed for
that development, but not for other purposes.
Then I might run Fedora and Rawhide, both, for development, and test
against both.
Just read this list and see how often Fedora blows up.
I think we should at this point try 'statistics for beginners'. You have
a list consisting of people who are heavily involved and people who join
because they have problems combined with a tendancy for those having
problems to be the ones who post (why email 'my server is working today'
to the list).
There are lies, damn lies, and statistics... I'd say the people on this
list are far, far better than average at fixing their own problems and
they still end up posting about things that they can't get to work.
Products such as RHEL are designed with a goal of avoiding regressions,
so what works continues to work and for many uses that is far more
preferable. To understand Fedora reliability you need to work out what to
measure. The end result of that is that while RHEL might be far less
likely to regress something that works well than Fedora, it is also far
less likely to cure something than Fedora.
I've only seen the 'cure' case where the hardware was newer than the
RHEL/Centos release that was loaded on it and fedora had a
correspondingly new driver. And updates for popular hardware are
usually backed into RHEL/Centos kernels.
--
Les Mikesell
lesmikesell@xxxxxxxxx
