Hi
Why is this question any harder to answer for fedora than for ubuntu?
And why, since it has been answered already, shouldn't approximately
the same answer be shared?
Not necessarily. Fedora isnt Ubuntu or viceversa though these two platforms have some similarities. If we are building exactly the same thing we might as well as toss out either of them. I have a draft document of what I consider new/unique features in the future FC4 release just to give you an idea
http://people.redhat.com/sundaram/fedora_notes.html
People building servers probably know what they are doing and will want a custom setup anyway.
I dont agree with you on this. Servers are usually consolidated. If you customise (I assume you mean rebuild RPMs here) you end up having to do it everytime a bug fix or security fix comes out
The case to make easy is the desktop
where the numbers are bigger and a lot of people are doing their
first install and don't have any way to make choices even if you
offer them.
Desktop numbers are far less than servers currently for Linux. I agree you with on the part about making it easier and not really providing confusing "choices" for the desktop class installation. Like I said before we are already moving in that direction. Nothing ground breaking here
I've always thought that if there were somewhere around 20 expertlysure. question is which 20. you got answers?. write them down in detail and post to fedora-devel. It would interesting to see the discusions evolving around that
chosen (and maintained) complete sets of programs already bundled
with descriptions of why you might want one set vs. another everyone
would be a lot better off than having to sort through 10,000 choices
at are just there because they are free.
That would help determine whats popular but would it be the only measure of what to bundle in Fedora Core. For example, FC4 will have Evince, a document viewer which is replacing gpdf. There is no way Fedora would have adopted SELinux or moved XMMS to Fedora extras based on stats. I could provide more examples if you want me to but I think you understand the other factors nowOne thing that would be interesting, but probably not currently possible to determine with current repositories, would be to find out how many times individual programs were requested specifically (that is, where a user made the decision to install a package rather than pulling an update because it happened to be included in a base package). Some metrics along these lines would really help in finding out what people are using.
regards Rahul
