Xfree or not Xfree? - Regression, progression, sanity, 3D and Xine

[Jim Cornette]

Mike A. Harris wrote:
> On Tue, 26 Aug 2003, Jim Cornette wrote:
> 
> 
>>>Downgrading metacity a few versions does not fix the problem.
>>>
>>>Downgrading XFree86-* to 4.3.0-18 does fix the problem.
>>>
>>>Looks like an XFree86 problem to me.
>>>
>>>Got your barf bag handy?
>>
>>Are the newer versions of XFree86 going to be regressing? Is there any 
>>sanity to versioning?
> 
> 
> This is an OS development cycle.  Development is occuring.  Bugs 
> get fixed, and problems get investigated.  Sometimes new problems 
> creep up, and then get fixed again later on.
> 
> If you can't handle that, then don't run the betas.  This isn't 
> meant for production usage by any longshot.  It's meant for 
> testing purposes by those who expect to find bugs, and can deal 
> with finding bugs, and reporting them so they can be fixed.

I expected bugs along the way. Plus, this is the first version of
Red Hat or any other version of Linux that will work on this particular 
computer.

I tried the latest version of the RH9 kernel and it locked up on pcmcia. 
Therefore, this is the best OS. Rawhide updating through Up2date is a 
great advantage. You get to see what all of these programs interactions 
cause on the OS. I'm glad that Xfree86 and GNOME are playing nice 
together again.

I was concerned that I would lose GNOME interaction again, if I upgraded 
XFree. I understand the balance with resolving the other users bugs. 
(Keyboard problems, especially)

> 
> People using the software, expecting it will be rock solid stable 
> when it is a beta, and then complaining does nobody any good 
> whatsoever.
> 

This is more related to ACPI being pulled because of instabilities. Then 
some users being inconvienienced by the lack of valuable features on 
their systems. It is meant to be more general. In case of regression for 
satisfying the major users, I wanted to be aware if this was going to 
happen, if I upgraded. It is more of not wanting to lose functionality 
again. If regression has to be tried to resolve problems, then I'd 
rather not bust the system again.

The versioning comment was more related to going from 1.1.x to 1.1.y 
might be stepping back in functionality. I see why you commented about 
the Abbott and Costello sounding explanation.

> 
> 
>>I'm holding off on updating any more Xfree86 versions for awhile
>>then.
> 
> 
> Please do.  However, you should note that if you wait until the 
> final version, and something has been fixed and is awaiting 
> confirmation that it's been fixed, and nobody tests the fix, then 
> it will likely remain broken.

Trial and error, before it goes to the masses, under beta or alpha is 
the only way to wring out the bugs. I see up2date and rawhide as a way 
to catch bugs early on. I opted out of the up2date upgrade and found the 
control center to be as a previous user explained. I only force 
installed redhat-release, after having trouble with rawhide-release not 
working with up2date. Regression is sometimes the better solution.


> 
> For the time being, if you want a stable, tested, and official
> product that is supported, use Red Hat Linux 9, Red Hat
> Enterprise Linux.  You definitely shouldn't be using a beta 
> release or rawhide packages however if you expect them to not 
> ever break.
> 
> Hope this clarifies things.
> 

It clarifies them better to me. Though I think I might have worded 
things which could be taken as wanting a perfect progressional upgrade 
cycle. I do, but do not expect it from a pre-QA'ed program set.

Jim

> 


-- 
If while you are in school, there is a shortage of qualified personnel
in a particular field, then by the time you graduate with the necessary
qualifications, that field's employment market is glutted.
		-- Marguerite Emmons