Casper's Life
As a follow up on my comments in “Five crucial things the Linux community doesn’t understand about the average computer userâ€, I saw a reference to this interview with Con Kolivas on /., who until recently was the maintainer of the -ck tree (linux kernel patch tree).
Even though people are already trying to bash each other with nifty comments about why, or why not Con is right in his statements.
There is one thing in this interview with stands out, and make the whole thing interesting. He did try, and he failed, for me (well I’m just a nobody) it looks like some of the kernel developers (sorry I’m not one of them, and I have to admit that most of them are very good at what they do) lack the perspective to be able to understand that they might not be doing the right thing – even if they actually are.
Many yeas ago I got my diploma as a software developer, and one of the things I learned doing that, is that the end user is always right (could also be the customer), that is as long as they do not want something which is obviously stupid.
Now we have a bunch of desktop users who would like to use GNU/Linux, but find that they a) lack support for their hardware b) they have the latest and brightest (which works), but still they don’t get the bang for their money. Well in that case, maybe the people who are responsible for what ever part of the kernel should have a look at it – yes the user might be wrong, and will be told off, but what if there is something wrong, can we afford that it does not get fixed?
In my humble opinion, we cannot afford to ignore the end user no matter how annoying it can be annoying to be told that ones code is rubbish by someone who doesn’t even know anything about programming……
If you enjoyed this article, please consider sharing it!
I agree with you in principle. This said, isn’t it also true that the kernel developpers are not really working for the end user? I mean they are obviously working first for the guys who are making the graphic interface and the various programs that the end user will use. So isn’t it more a problem that should concern these guys first?
OpenSource have it’s positive sides, and negative sides. First Linus had a great idea when he started the kernel back in the early 90′s, and Richard Stallman had another great idea when he defined GNU (Gnu Not Unix), which is fsf (free software foundation) today, with the kernel and GNU we today have GNU/Linux which is the right way of describing a Linux distribution.
Now back to the matter at hand, the issue is that OpenSource development could be the equivalent of taking a handful of gravel and throw it in the water, and see what happens. There is a huge amount of projects which have different leadership(s), and as with most things which are done for free (in many cases that is), then it comes down to arrogance (sorry if I just steeped on some feet), or self important or what ever word is the right fitting. I’ve been there, you want to make sure that everyone know about what you’re doing, etc. With OpenSource it becomes even more clearly, as no one does this without trying to get something out of it (there is no better motivation than getting famous), and with all the things which goes into the kernel there will be people who either are for or against changes.
In the case of Con Kolivas, yes he might have done things wrong, but he did a lot of things right, as for the decission Linus did regarding the scheduler (there is an quite interesting read on kernel trap here), well we all know that Linus sometimes have an issue in dealing with people, and the way he communicates can put people off.
As for, what makes these things go wrong, first the kernel is centered around a getting things to work (Linus complained earlier this year that people submit more bug in new code than they fix), and for the scheduler it’s (prior to the latest installment), was based on an old design which could be called “fair use” everyone get the same amount of time on the CPU, now this is great if you have the latest and brightest hardware of if you’re running a server (servers are easier to write code for….. yes, yes I know that is not always right), but for a desktop…
On the desktop you have to decide what is important, is moving a window (I btw. hate waggly windows), or is that thing which checks your email while you’re playing your favorite game more important, or is actually the game more important, and then how do you manage this….
So yes they are concerned, but I doubt that anyone actually know how to solve that issue, except that Con actually got quite close to find a way to do it, the scheduler could be compared to higher math (the stuff you get a PhD for).
That is the kernel, then we have GUI, which can come in two different ways (there are more but…), you will either be to Gnome, or to KDE, they are both great, which is why people are fighting the great GUI war over what is best… The GUI just sit there, and hope for the kernel to give it what it want, and there is not much they can do about that, so back to the kernel developer…..
The kernel developers rather like to implement new stuff (more fun – trust me), they do not want to spend weeks (months or years) on end trying to figure out why process A is slower than process B, even though it shouldn’t, that is boring (troubleshooting is something which pays big bucks if you’re good at it), which is why it’s more fun to play around with new exiting stuff, than trying to figure out why something doesn’t really work as designed.
Yes I rant, and I know that people get annoyed, but never expect that GNU/Linux will throw Microsoft Windows off the home-desktop as long as it is a combersom to use as it is. Corporates have a gain on this, as they can save huge amounts of money, and with a locked-down workstation they can control it, and then it suddenly gets possible, but the home-user will continue to use MS Windows as long as GNU/Linux is not a a viable solutions – ie. read play games, etc….