Left by Rawhide

Posted Jul 16, 2012 17:10 UTC (Mon) by louie (guest, #3285)
Parent article: Left by Rawhide

If the current form of Rawhide better suits the project's needs and leads to better releases, then changing Rawhide was the right thing for the project to do.

It only suits the project's needs well because the project, collectively, is a total failure on the QA front, so what's one more source of failure and wasted opportunity. This is not a new problem. My first publicly recorded mini-rant on the subject is from 2007, and it was already a long-standing problem at that time. As I said then:

Let me be clear- I feel that quality is one of the biggest possible advantages free software can have over proprietary software, specifically because you can have hundreds or thousands of people testing the latest code every day. If you're not taking advantage of that, you're throwing away one of the biggest advantages we have over proprietary software.

(emphasis added)

With a useless Rawhide, Fedora continues to throw away that potential QA advantage. But that's nothing new.

to post comments

Left by Rawhide

Posted Jul 16, 2012 17:23 UTC (Mon) by dowdle (subscriber, #659) [Link] (3 responses)

louie, So given your dismal opinion of Fedora / Rawhide, what have you moved on to? What distro is the most stable? ...staying somewhat bleeding edge for desktop stuff? ...doing a lot of original development work on many small and some major packages? I'm guessing the answer is different for each question but maybe not.

Fedora definitely isn't for everyone but I prefer it and refer to it as the "innovator distro". We all have our preferences and that is one reason I'm glad so many distros exist.

People sometimes freak out about the number of Linux distros but I ask them why aren't there multiple flavors of Windows and Mac available... besides the few releases provided by their makers? The answer is obvious, because they don't allow third parties to remix and release like Linux does... but you have to wonder just how many Windows there would be and how many Mac OS X's there would be if people were allowed to remix it. My guess is that they would have the same "problem" we do if only they were allowing it.

Left by Rawhide

Posted Jul 16, 2012 21:17 UTC (Mon) by louie (guest, #3285) [Link] (1 responses)

I still use Fedora (not Rawhide, which I would prefer to use). Fedora itself is a perfectly competent distro, with both practical and moral aims that work for me. I just think that treating Rawhide as a dumping ground squanders an opportunity to much better achieve those practical/pragmatic aims.

Left by Rawhide

Posted Jul 18, 2012 3:32 UTC (Wed) by AngryChris (subscriber, #74783) [Link]

This is where i'm at right now. I swap between Debian and Fedora every couple of years. What amazes me is how good Fedora is when it appears (to me) to be so grossly mismanaged. I don't see how Rawhide can be remotely useful for "seeing how things work together" if no one is running it (i.e.; making those things work together). As recently as last year, the policy board mailing list was overrun with talk of mission statements. It's like watching Initech (Office Space) actually publishing great software.

I have a feeling that most good Free software is good in spite of how the project is run rather than because of it (for any given Free software project).

Left by Rawhide

Posted Jul 17, 2012 9:12 UTC (Tue) by viiru (subscriber, #53129) [Link]

> People sometimes freak out about the number of Linux distros but I ask
> them why aren't there multiple flavors of Windows and Mac available...
> besides the few releases provided by their makers? The answer is obvious,
> because they don't allow third parties to remix and release like Linux
> does... but you have to wonder just how many Windows there would be and
> how many Mac OS X's there would be if people were allowed to remix it. My
> guess is that they would have the same "problem" we do if only they were
> allowing it.

There are already nearly endless amounts of custom Windows installer cd images, with different sets of updates and additional drivers added. Distributing those is of course illegal, but since there is a need (there is plenty of hardware where the standard images wont even boot, many manufacturers no longer provide install disks and even if they did most users lose those within the first week of ownership) there is a service.

Left by Rawhide

Posted Jul 16, 2012 21:22 UTC (Mon) by johannbg (guest, #65743) [Link] (5 responses)

"It only suits the project's needs well because the project, collectively, is a total failure on the QA front"

Care to elaborate on that comment?

Btw without proper test cases and without proper debugging procedures hundreds or thousands of people testing the latest code every day is meaningless...

Left by Rawhide

Posted Jul 16, 2012 21:29 UTC (Mon) by louie (guest, #3285) [Link] (3 responses)

Btw without proper test cases and without proper debugging procedures hundreds or thousands of people testing the latest code every day is meaningless...

People testing the code gives you data on the parts of the software that people actually use. That's certainly not perfect, and shouldn't be the end-all/be-all of testing, but having people use and report back on the parts that get used is absolutely the best software testing methodology there is, period.

Left by Rawhide

Posted Jul 16, 2012 21:34 UTC (Mon) by johannbg (guest, #65743) [Link] (2 responses)

And this is relevant to my comment how?

Left by Rawhide

Posted Jul 16, 2012 21:53 UTC (Mon) by louie (guest, #3285) [Link] (1 responses)

hundreds or thousands of people testing the latest code every day is meaningless...

Hundreds or thousands of people testing the code is never, ever meaningless. If you think it is, your QA processes are broken.

Context?

Posted Jul 23, 2012 11:06 UTC (Mon) by gmatht (subscriber, #58961) [Link]

I am not sure what the precise context is here. But if the context is having hundreds of thousands of people test software that is known not to work at all, then you aren't going to get much useful feedback. In many projects there is a "latest software that passed basic automated tests" tree, which may be subtly broken but will at least partially work. I don't see any point in having hundreds of thousands of people test anything newer than that; having a thousand "nothing works" bug reports show up at the same time doesn't really help anyone.

Left by Rawhide

Posted Jul 17, 2012 0:02 UTC (Tue) by rgmoore (✭ supporter ✭, #75) [Link]

Users and test cases are different and provide different kinds of information. Test cases are great because they provide reliable, high quality information, but they're limited; nobody has enough resources to test every configuration their users are going to try out. User reports are messy, but they're documentation of real problems that your QA has to fix. The continued existence of user error reports is proof that the kind of disciplined testing represented by proper test cases and debugging procedures is inadequate. If it were enough, no problems would ever get through for users to find them.