The "Fedora" mailing list

[bill at billhorne.homelinux.org]

TWIMC:

RedHat's "Fedora" project has a mailing list.

The RedHat developers are having a great time.

FWIW.

Bill

From: "Mike A. Harris" <mharris at redhat.com>
To: fedora-list at redhat.com
Subject: Re: Why two GCCs in FC1??
Organization: Red Hat Inc.
X-Unexpected-Header: The Spanish Inquisition
MIME-Version: 1.0

On Thu, 6 Nov 2003, Alexander Grekhov wrote:

>> This has been used heavily all over the kernel and although it has
>> been cleaned in many places in 2.4.x kernels, it is still present
>> in several places.
>> 2.6.x kernels should build with GCC 3.3.x just fine.
>
>Thanks for the explanation. Still I think it would be less confusing and 
>more logical to release 2.4.x-based FC1 with GCC 3.2.x and move on to 
>kernel 2.6.x and GCC 3.3.x in FC2.

That is not far more logical.  That would throw away 6 months of 
compiler technology development and improvements if not more, and 
delay them from being widely used by people for another 3-6 
months or more.  There is no valid good technical reason to do 
that.

Fedora Core is for shipping new technology, and that is what we 
plan to do.  In some cases, this means at least for the 
compiler/kernel that multiple compilers will be needed, and that 
is just going to be a fact of life from time to time, so people 
will have to get used to it one way or another.  This isn't 
something new either.  All Linux distributions do this, and Red 
Hat is no exception.  We don't want to wait 6 months to use 
tonnes of performance enhancements and other compiler benefits 
because some people can't figure out how to compile their kernel.

Read documentation, ask questions, learn more about the system.  
If you need to compile a kernel - which the massive majority of 
users never ever do, then you are in the small minority of people 
who compile their own kernel, and you, as a hobbyist need to 
learn how the system works.  Compiling a kernel is a developer 
task, and that means you learn more if you want to do it.

Holding back evolving technology because it makes a simple small 
task like compiling a kernel, which is a developer thing, 
more difficult for non-developers, is not something we're 
ever going to consider a favourable or good thing.  We pay very 
talented people to develop new compiler technology, and that is 
not just done for Red Hat, that goes directly into official gcc.  
And we fully plan on using those improvements in our Fedora Core 
distribution as soon as they are stable and we can possibly 
include them in the distribution.

The kernel on the other hand, is very picky about the compiler, 
and new compilers almost always change many assumptions the 
kernel source code has about how the compiler works.  This almost 
always means with every new major compiler release, the kernel 
will need to be compiled with older compilers until kernel folk 
are able to completely update the entire kernel to work with the 
new compiler, a task that potentially takes a very long time to 
do, and to beta test, etc.  That time period is usually longer 
than the time period in between distribution releases.

So, while your frustration is registered and noted, your solution 
is not viable, and is not in line with the Fedora Project's 
stated goals of using new technology.

"I can't figure out how to compile my kernel" is not a valid 
reason to throw away perfectly stable working new technology.


-- 
Mike A. Harris     ftp://people.redhat.com/mharris
OS Systems Engineer - XFree86 maintainer - Red Hat


--
fedora-list mailing list
fedora-list-request at redhat.com?Subject=subscribe
http://www.redhat.com/mailman/listinfo/fedora-list