[K12OSN] *EARLY* Alpha of K12LTSP 3.2

[Andrew]

It isdn't so much redhat improving over the original author--it is a 
matter of a series of distro specific bug reports that help get the 
problem ironed out at both ends.
I stick w/ debian testing at home btw, and don't mix stable/unsrtable 
etc.  I gotta ask though, what is it you think can be done  (in a 
production sense, not in a running the latest subversion of something 
sense) on fedora that can't be done on whitebox? Certainly not much 
multimedia..... Not much by way of failover clustering... what then? 
They just aren't that far apart yet.  The biggest improvement made in 
Fedora, the inclusion of kernel 2.6, is useless in a k12ltsp environment 
so far, as k12 ltsp.org supports up to kernel 2.4.26 as of today.
All the recent nifty improvements in the way linux does things really 
hang on kernel 2.6. Whitebox uses kernel 2.4.25 (modied, of course) do 
version 2.4.25rhel and kernel 2.4.6rhfc2 seem that far apart to you in 
funcionality?  Sadly openmosix also seems to still like 2.4 kernels.  So 
long as our networking  and clustering stuff depends on , as you  say, 
an ancient kernel, it seems kinda silly to be calling whitebox ancient 
and obsolete.
-AF

Les Mikesell wrote:

>
>Yes, but Redhat develops a tiny fraction of a Linux distribution.
>Most of the code comes from thousands of other developers
>who are learning as they go, fixing old bugs with each
>new version and improving their work.  Redhat promises to
>backport the serious bugfixes into old releases for the
>support lifetime of a distribution (and you can see why
>this would kill them before they changed the versioning
>model).  But, do you really think RH can do a better job
>of making programs work correctly than the original
>author?  I'll take one I know about as an example: perl
>has some serious bugs that keep it from working with
>an application I use unless it is 5.8.3 or newer, which
>you find in fedora core 1 and later.   RHEL has something
>earlier and apparently considers handling different character
>sets correctly a 'feature' so they don't backport the fix. 
>
>  
>