Interview with Con Kolivas...

The_ouroboros · 2007-07-25 10:22:06

If there is any one big problem with kernel development and Linux it is the complete disconnection of the development process from normal users. You know, the ones who constitute 99.9% of the Linux user base.

[REF: http://apcmag.com/6759/interview_with_c … is_boring]

So the developer said....
What you think about it?

Tnks

Last edited by The_ouroboros (2007-07-25 10:23:37)

oli · 2007-07-25 13:51:26

It's a pity, if you see such an capable man leaving the kernel development.

Kardell · 2007-08-04 23:34:43

It`s shame on Linus, Morton & Molnar. They want to monopoly their kernel as the closed dev order.
They even didn`t give him an alternative, just let him go.
Why, cause he does not work for Red Hat? It looks that they were ignoring him all the time and favouring Molnar.

Last edited by Kardell (2007-08-06 07:54:16)

sykesm · 2007-08-05 09:21:35

thanks for posting the link. A good read; it will be interesting to see where things go over the next 35 years, the last thirty five have certainly had their moments. Is the desktop experience going to be reduced to a Web browser, and a box which you chuck away every two or three years; you do not know or care what the OS is. I suppose from most desktop users point of view, the ideal situation.

chaosgeisterchen · 2007-08-05 15:55:22

It's frightening to read, that most of the Linux development is done so extremely non-desktop-centric. Given that most Linux users are desktop users, Linux should have a heavy emphasis on improving desktop performance - the only way to get way past Windows. They are wasting their chance to become a major force in desktop business, unfortunately.

Mikko777 · 2007-08-05 17:00:01

Oh please, latest news is that Con is a troll.

Linus already put him in his place (explained why con didnt manage to get his code merged)
and that the linux developers dont care about desktop is also bs because, surprisingly thats what most of the devs end up using....

Too lazy to find the link, you'll get the point though.

chaosgeisterchen · 2007-08-05 17:48:21

Probably the article you meant.

Well, two opinions, two persons. I think it's a fact that Linux is more compatible concerning hardware and tries to support everything but by doing so, it is not able to focus on improvements concerning a certain platform and hardware constellation like Apple is doing.

mcv · 2007-08-06 08:33:58

What I don't like in Linux are (among others):
- Memory allocation, where allocator will not tell you that there's insufficient memory, but will use OOM killer to kill Your precious RDBMS (for example),
- disability to force unmounting filesystem (as in FreeBSD for example), where in most cases I have to "-9" some applications (Konqueror for example sometimes leaves using dev althoug no Konqueror window is currently opened).
- very bad responsivity, when system is copying some big files. (Well, maybe this responsivity isn't bad on quad-xeon with 4G RAM and 15000RPM Cheetach HDDs (maybe developers have such machines), but it is on my Athlon-XP 2800, 768M RAM and 5400 RPM HDD.) There goes the same, when system is swapping memory… in the end data is being copied using HDD.

I agree that kernel shouldn't be moved towards desktop only, but why did Linus reject patch for changing cpu scheduler on-the-fly? Because user would have too much control? That's ridiculous. Linus pisses me off for some time…

retsaw · 2007-08-06 08:57:26

mcv wrote:

I agree that kernel shouldn't be moved towards desktop only, but why did Linus reject patch for changing cpu scheduler on-the-fly? Because user would have too much control? That's ridiculous. Linus pisses me off for some time…

Did you not read his reasons? Basically it was something like if there were two schedulers that had problems with different loads, the user would just switch to the other scheduler if one had a problem with the load. This would lead to problems getting ignored and development being split across two schedulers. Linus wants a single scheduler that works well with any load, so a pluggable scheduler system is counter-productive to his goals.

It may not necessarily be the best choice, but is a valid reason for his decision.

jb · 2007-08-07 14:41:25

I still don't get that argument. Of course some people will prefer one scheduler to another; they are meant for different workload types. If there's a server-optimized and desktop-optimized scheduler, yet one is busted to the point it causes everyone to use the other, then it will either be fixed or deprecated... I thought that was one of the up sides to OSS.

And as for the 'split development' argument... why he would use that. There will be people interested in working on one scheduler, and those who would prefer to work on the other. I doubt there's many kernel devs now that decide to work on something by picking a name from a hat. They will find something they are either interested in or paid to be interested in and work on that. For those that do decide to 'switch camps' and work on the other, really it just means that they find the other more interesting and/or relevant. But really, does he expect us to believe the Kernel dev team would suddenly be short-handed due to adding one more pluggable system?

...but that's just my take on it.

The_ouroboros · 2007-08-07 16:33:21

good point jb... I agree with you..

Kardell · 2007-08-07 21:45:46

jb wrote:

I still don't get that argument. Of course some people will prefer one scheduler to another; they are meant for different workload types. If there's a server-optimized and desktop-optimized scheduler, yet one is busted to the point it causes everyone to use the other, then it will either be fixed or deprecated... I thought that was one of the up sides to OSS.
And as for the 'split development' argument... why he would use that. There will be people interested in working on one scheduler, and those who would prefer to work on the other. I doubt there's many kernel devs now that decide to work on something by picking a name from a hat. They will find something they are either interested in or paid to be interested in and work on that. For those that do decide to 'switch camps' and work on the other, really it just means that they find the other more interesting and/or relevant. But really, does he expect us to believe the Kernel dev team would suddenly be short-handed due to adding one more pluggable system?
...but that's just my take on it.

Exactly
I thought that the unusual GNU devs - users symbiosis works in different way.
But now I see that we`re becoming only a pack of testers that serve for commercial Linux branches, solutions for big companies, Red Hat etc.
I don`t see the point, why we still should test their high-end solutions on our desktops. It`s time for Linux of 21 century, not Linux only for guys with long beardes. ;-)

Arch Linux

#1 2007-07-25 10:22:06

Interview with Con Kolivas...

#2 2007-07-25 13:51:26

Re: Interview with Con Kolivas...

#3 2007-08-04 23:34:43

Re: Interview with Con Kolivas...

#4 2007-08-05 09:21:35

Re: Interview with Con Kolivas...

#5 2007-08-05 15:55:22

Re: Interview with Con Kolivas...

#6 2007-08-05 17:00:01

Re: Interview with Con Kolivas...

#7 2007-08-05 17:48:21

Re: Interview with Con Kolivas...

#8 2007-08-06 08:33:58

Re: Interview with Con Kolivas...

#9 2007-08-06 08:57:26

Re: Interview with Con Kolivas...

#10 2007-08-07 14:41:25

Re: Interview with Con Kolivas...

#11 2007-08-07 16:33:21

Re: Interview with Con Kolivas...

#12 2007-08-07 21:45:46

Re: Interview with Con Kolivas...

Board footer