On Wednesday 29 October 2003 02:36 pm, robin wrote: > Since when is a stable release "bleeding edge"? Bleeding edge is > Cooker. Bleeding edge is running alpha apps. Bleeding edge is using a > development kernel.
Mandrake Linux is considered much more cutting edge and state of the art than many other distributions including Debian or Red Hat. If one valued stability over getting the latest technology, there are distributions that provide that. I don't remember any Mandrake release that wasn't followed by several rounds of patches to fix problems. If you were running on the assumption that a new release was stable, then perhaps my cynicism has just been protecting me. I think that any new release, especially one that was tested with a fairly limited audience and on probably limited platforms should be considered bleeding edge. This is especially true where the test platform is probably not the lowest level of computer equipment. Most people running cooker or even installing the release candidates are not your average test bed so common sense should indicate some caution before installing and using a brand new release of software. Given the fact that it is still not available to persons not members of Mandrake Club, I would argue that the release pool is still too small to be considered to be fully vetted by anyone. > > It is not outrage, it is simply annoyance that there are so many more > > people prepared to bitch and look a gift horse in the mouth than there > > are people prepared to contribute and offer support, or accept personal > > responsibility for knowing about their own computer equipment. Criticism > > of something that should not have been done is one thing, blaming a linux > > developer because he wasn't smart enough to protect an idiot designer at > > LG from his own stupidity is something else entirely. > > You are making an assumption that the people who complain (or "bitch", > as you so charitably put it) are a separate group from the people who > contribute and offer support, or for that matter pay for their software. I am making no assumptions. You said that the kernel developers should have tested the patch with LG CDROMS and unless you want to admit to chiming in with "hindsight is 20/20" advice, I would argue that suggesting such testing would also involve several varieties of motherboards, hard drive controllers, network cards, modems, CD Burners, DVD Burners and the list goes on. In each case, it is possible that some change introduced into the kernel will have some negative effect on legacy hardware. The only way to be sure is to test exhaustively. That means loading them up, configuring the software and then testing each one in turn. That takes time and resources, all from people who have regular jobs, in some cases that don't involve Linux. MS, which has a much bigger QA group and loads more resources than does probably any Linux developer does NOT do this type of testing. They require hardware manufacturers to do it themselves. Still think that it is trivial for Linux developers to do it? Would you pay the extra cost to have it done? And wait the extra time for the release? If I were willing to wait the extra time, I would still be running Debian, that is a very stable and well tested distribution. Testing will add on to the release schedule and extend the period when releases are made and probably reduce revenue for a company that is already not on the best financial ground. And this for software changes that Mandrake itself may not be making or even familiar with. The point that I am making is that short of that hindsight advice, you are suggesting that they create and maintain a test lab of machines and hardware for testing purposes but not suggesting how cash-strapped developers are supposed to pay for such a lab. Thus my characterization about people complaining but not offering solutions. I think it is apt. > > At the point that MS starts to deploy closed box console computers to the > > general public, I am sure that they are going to point to this particular > > incident to explain why it is preferable for people to not install their > > own OS or software, and not to buy separate computer components. > > That was exactly my point. And mine that followed. The only way to be sure is to operate with a closed box. No developer or company will take responsibility for something that they have not personally vetted or built. If it is impossible for Linux developers to exhaustively test each piece of hardware, then that responsibility falls to those of us who install the software at our own risk. If anyone doesn't like taking that responsibility themselves, there are plenty of others who will take it for you, you simply have to pay for that. Or wait for others to do so. > > Unlike some others in the community, I am not at war with MS, I do not > > seek the destruction of MS, I am not on a holy crusade, nor am I an > > evangelical pushing a philosophy. I am simply someone who enjoys the > > fruits of labor that have been given to me and tries to contribute back > > when I am able to do so. > > With the implication that those who criticise don't do this, perhaps. Nothing at all implied and you missed my point. I am not encouraging Linux because I want to defeat MS and I don't care and think that the Linux community should not care one whit if someone wants to install MS software because the Linux community failed to live up to their expectations and demands about providing the type of product that they wanted. I don't care if we lose some fair-weather friends to Windows because their crappy drive blew up due to a hardware design screw-up. My point was, I don't care about MS's monopoly, or anyone else's for that matter because I still have choices and the only cost to me has been time and effort to learn how to do for what I need for myself or pay or find someone else to what I can't. So if this little glitch drives some other people to Windows, I say, don't let the door hit you on your ass on your way out. > > If the price of getting someone to use Linux is that I have to dummy it > > down to the point where I replicate the same mistakes, bad design and > > idiotic marketing bullshit that MS has propagated, then I say that price > > is too high. > > I don't see what any of this has to do with dummying down Linux. > Besides, would you rather go back to the days when installing Linux > meant editing configuration files in emacs? No, but I don't think that the answer is to implement the same design mistakes trying to dummy proof software for someone that doesn't want to accept personal responsibility for learning and figuring out some things for themselves. That includes buying cheap-ass, defective hardware and then complaining because the Linux OS, performing a completely standard activity caused said defective hardware to blow up. > > I wouldn't consider trying to perform electrical design in my home or car > > because I am not qualified to do so. Buying a particular OS does not > > make someone qualified to configure or maintain a computer. If someone > > wants something that is just like Windows, I say, let them use Windows. > > And if someone doesn't want a user-friendly Linux, I say, let them use > Debian. But as I said, this isn't the issue here. No, it is not about user-friendly. It is about compatibility, logos, certification and expectations. It is about either checking the hardware out with the manufacturer for support or checking with others who have looked at it first. Either that or you take your chances, throw the dice, and don't whine when it comes up craps. > > No, but asking developers to keep fully stocked hardware labs for testing > > purposes is hardly a simple criticism. You do not provide any mechanism > > for them to easily implement your suggestion and leave the onus of doing > > so entirely on them. Not to mention the extra time and resources > > required to conduct such testing. I do have some knowledge of what is > > involved and it is NOT trivial. > > If LG were some really obscure hardware manufacuter, I would agree. The > fact is that these are some of the most common CDROMs on the market. And > as someone pointed out, they cost around $12. Again, hindsight is 20/20. Now, if you had made this observation about testing against LG drives before the release, I might have a tad more respect for the warning. Since you are suggesting this after the problem has been found, I am giving it what I think it warrants. Anyone can chime in after the fact and talk about what someone else should have done. It does not sound helpful and, IMO, pisses off people who volunteer a lot of time to benefit others because it sounds and awful lot like the blame game. Try making suggestions before the problem crops up and people will probably be much more receptive. If I sound a little emotional here, it is because I do testing for a living and I constantly listen to people who stroll in after the fact and say very matter of factly, "well why didn't you test this part where the bug was?" and then look around as if we are all supposed to be awed by their brilliance. Gee, why didn't I think of simply testing where the bugs are? Well, I guess it has to do with not keeping my subscription to the Psychic Friends Network current so I don't know where the bugs are. I have found through practice that figuring out where to look for problems before they are found is a tad more difficult than pointing them out afterward. At this point, LG probably won't have much business in the immediate future because even if the Linux glitch didn't burn them enough, the subsequent exploits probably will. So, more than likely, this particular problem won't crop up again because any hardware manufacturer who is left is going to be wary of finding themselves in a similar position. Since the problem with LG drives is least likely to be repeated and you are so anxious to tell the developers how important it is to test for possible glitches, perhaps you want to make a prediction as to the next bug or problem that will come up so that they can be sure to test for it. If I sound a little offensive here, I am sorry but again, I think that the point that I am making is very appropriate. People who don't understand the complexity of QA often completely underestimate how hard and resource intensive it can be. And doing it halfway is almost worse than not doing it at all, at least if you just don't do it, you are not fooling yourself thinking that you have found the worse stuff and giving your customers false expectations about what to expect. Doing it right takes a lot of time, people and money. Short of suggesting where to find those resources, suggesting to developers that they should have tested a particular thing where a bug just happened to crop up serves absolutely no point at all. It does however, yank my chain. > > For those that wanted to gain the benefits, the only thing that they > > needed to do was actually WAIT before installing until others had had a > > chance to do the same type of testing that you propose. > > As I said, it was a stable release. That should mean testing was > complete. Not testing on every obscure bit of hardware on the planet, > just the most common stuff. Fine. Define common and give me a list because that is certainly something that I could use in my current line of work since I test for a living. Oh, and points off if your list doesn't match the fifty others that are submitted all by people who have different ideas as to what is the most common hardware set built completely according to what they themselves own. -- Bryan Phinney Software Test Engineer
Want to buy your Pack or Services from MandrakeSoft? Go to http://www.mandrakestore.com