Views from an Red Hat -> Ubuntu -> Fedora migrator

[Stephen Gordon]

On Mon, Oct 22, 2012 at 3:20 PM, Lennart Sorensen
<lsorense-1wCw9BSqJbv44Nm34jS7GywD8/FfD2ys at public.gmane.org> wrote:
> On Mon, Oct 22, 2012 at 02:54:08PM -0400, D. Hugh Redelmeier wrote:
>> I don't remember it that way.  I admit that my memory isn't 100%
>> reliable.
>>
>> I think that this was the time where there were GCC project political
>> problems.  It was important to move GCC forward -- it was an important
>> pillar of free software.  FSF's GCC was stalled, so there was a fork.
>> Eventually, the fork won.  I think that Red Hat was pushing on this,
>> ahead of the other distros.
>
> http://gcc.gnu.org/gcc-2.96.html
>
> That was one of the problem with redhat 7.x
>
> They were too impatient to wait for 3.0 to be done.  It was far from
> stalled at the time.
>
> But the decision makers at redhat just had to have the new gcc 3.0
> features that weren't quite ready yet.  Never mind what the gcc developers
> think.

Of course there are two sides to every story:

http://www.redhat.com/advice/speaks_gcc.html

>> The only problem that I remember is that there were kernel bugs that
>> only showed up with the new compilers so you needed to have the old
>> compilers around if you wished to compile the kernel.
>>
>> The Linux kernel isn't really compiler-independant, and was less so
>> then.  I consider this an unfortunate flaw but Linus doesn't.
>>
>> I'm not a C++ guy (an understatement) but my impression is that g++
>> has been embarassingly non-conformant and as it improves, more code
>> written for it breaks.  Perhaps that is what you are talking about.
>>
>> Summary: users experienced problems, but these were on account of
>> progress that Red Hat was pushing.  I consider it a feather in their
>> cap.
>
> If the gcc developers say "Don't do that", then that is NOT pushing
> development.  It is rather hindering it.  You are wasting their time
> with issues they don't want to deal with.  gcc 3.0 was getting close
> to done.

Some gcc developers agreed with that line of thinking, some didn't.
Acting as if GCC community universally agreed it was a Bad Thing TM is
revisionist.

    http://lwn.net/2000/1005/a/rh-tools.php3

Steve
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists