Linux kernel development going too fast?

Linux kernel development going too fast?

Post by Vassilis Pand » Fri, 31 Aug 2001 22:04:37



First of all don't get me wrong. I like the way Linux kernels are
released , and above all , the speed in which security exploits are
fixed. But , aren't we going a little too fast? It seems to me that
Linux kernels are released too fast for someone to cope with. Unless
you have a lot of time to waste , you cannot keep up to date with the
kernel. I think that kernels should be released slightly slower ( i am
talking about patch versions).This will allow developers to test their
code a little more ,and before the kernel is deemed "stable" , fix the
very obvious bugs. Real rapid kernel changes should normally only
happen when there is a security exploit. Then a new kernel can be
released ,without having anything more than the previous one; just the
fixed security exploit. Then stable kernels will be stable , and
development development.
 
 
 

Linux kernel development going too fast?

Post by Grant Edwar » Fri, 31 Aug 2001 23:32:41



> First of all don't get me wrong. I like the way Linux kernels
> are released , and above all , the speed in which security
> exploits are fixed. But , aren't we going a little too fast? It
> seems to me that Linux kernels are released too fast for
> someone to cope with.

Then don't try to keep up.  I'm still running 2.2.16, and will
probably continue to do so until releases of 2.4.x slow down.
Nobody says you've got to update every time a new kernel patch
comes out.  If your systems does the work it's supposed to,
leave it alone. ;)

Quote:> Unless you have a lot of time to waste , you cannot keep up to
> date with the kernel.

Why do you need to?

Quote:> I think that kernels should be released slightly slower ( i am
> talking about patch versions).

I think that the sooner the code is out there for people to
play with, the sooner the bugs get fixed.  If you don't want to
be part of the bug-fixing process, then don't update every time
a new patch comes out.

--
Grant Edwards                   grante             Yow!  .. Now KEN and BARBIE
                                  at               are PERMANENTLY *ED to
                               visi.com            MIND-ALTERING *...

 
 
 

Linux kernel development going too fast?

Post by Dave Pla » Sat, 01 Sep 2001 02:51:31


Quote:>First of all don't get me wrong. I like the way Linux kernels are
>released , and above all , the speed in which security exploits are
>fixed. But , aren't we going a little too fast? It seems to me that
>Linux kernels are released too fast for someone to cope with. Unless
>you have a lot of time to waste , you cannot keep up to date with the
>kernel.  I think that kernels should be released slightly slower ( i am
>talking about patch versions).This will allow developers to test their
>code a little more ,and before the kernel is deemed "stable" , fix the
>very obvious bugs.  Real rapid kernel changes should normally only
>happen when there is a security exploit.

That's the way things _already_ seem to take place in the later stages
of a stable kernel series development cycle.  The 2.2 kernels, for
example, are being updated only every 2-3 months.

In the early days of "stability" of a new kernel series (say,
2.$even,[0-20]) you can expect to see new patch releases occur fairly
frequently.  This is, I believe, due to a need to reach a reasonable
level of stability across a wide range of platforms.

In theory, a new "experimental" kernel series would not be snapped
over to the "stable" status until it was truly stable.  I don't think
that has ever happened for Linux, and suspect it's rather rare for any
operating system.  The reason is simple - for as long as a kernel
series is listed as "experimental", only a relatively small number of
people will be willing to use it on a regular basis.

After much hard work, such a series may be declared to be "stable, as
best as we can tell given the set of people and systems doing the
testing", and then snapped over to become the "dot-zero" release of a
new stable kernel series.  Once this happens, the number of people
using it jumps by a VERY large factor - probably 100:1 within a few
weeks - and the diversity of systems using it increases by a large
factor as well.

Immediately thereafter, there's always a rash of new problems
discovered, and a flurry of patches to truly "stabilize" the truly
stable kernel.  This "production shakedown" period tends to last for
several months.

If you want true, long-term stability, wait a while longer before
jumping on the "latest and greatest thing" bandwagon.  You really have
your choice of three paradigms - the older stable kernel series (e.g.
2.2), the newer and not quite as stable series (2.4), or the
bleeding-edge experimental series (2.5).  Ya pays your money and ya
takes your choice (and chances!)

--

Visit the Jade Warrior home page:  http://www.radagast.org/jade-warrior/
  I do _not_ wish to receive unsolicited commercial email, and I will
     boycott any company which has the gall to send me such ads!

 
 
 

Linux kernel development going too fast?

Post by Anonymou » Sat, 01 Sep 2001 02:52:28


No it can never go too fast.
You don't have to install new versions.
This is one of the great things about open source.

I only upgrade the kernel if there is an official
RedHat update.  This way I always get a kernel version
that at least has gone through a limited QA at RedHat.
It saves me the trouble of having to do it myself.

With proprietary software, you can be stuck in a situation
where you have to upgrade because some old software
won't run.  But with Linux, you can take old or new
software and just recompile it, most of the time,
you don't even have to do this step.


> First of all don't get me wrong. I like the way Linux kernels are
> released , and above all , the speed in which security exploits are
> fixed. But , aren't we going a little too fast? It seems to me that
> Linux kernels are released too fast for someone to cope with. Unless
> you have a lot of time to waste , you cannot keep up to date with the
> kernel. I think that kernels should be released slightly slower ( i am
> talking about patch versions).This will allow developers to test their
> code a little more ,and before the kernel is deemed "stable" , fix the
> very obvious bugs. Real rapid kernel changes should normally only
> happen when there is a security exploit. Then a new kernel can be
> released ,without having anything more than the previous one; just the
> fixed security exploit. Then stable kernels will be stable , and
> development development.

-----=  Posted via Newsfeeds.Com, Uncensored Usenet News  =-----
http://www.newsfeeds.com - The #1 Newsgroup Service in the World!
 Check out our new Unlimited Server. No Download or Time Limits!
-----==  Over 80,000 Newsgroups - 19 Different Servers!  ==-----
 
 
 

Linux kernel development going too fast?

Post by The.Central.Scrutinizer.nos.. » Sat, 01 Sep 2001 02:58:44



>If you want true, long-term stability, wait a while longer before
>jumping on the "latest and greatest thing" bandwagon.  You really have
>your choice of three paradigms - the older stable kernel series (e.g.
>2.2), the newer and not quite as stable series (2.4), or the
>bleeding-edge experimental series (2.5).  Ya pays your money and ya
>takes your choice (and chances!)

I've found the 2.4 series (I'm currently running 2.4.4) to be extremely fast
and reliable.  2.4.4 runs so well that I'm really not interested in upgrading
to a newer subversion.
 
 
 

Linux kernel development going too fast?

Post by Dietrich Schaffe » Sat, 01 Sep 2001 03:38:21




>> First of all don't get me wrong. I like the way Linux kernels
>> are released , and above all , the speed in which security
>> exploits are fixed. But , aren't we going a little too fast?

[...]

Quote:> Nobody says you've got to update every time a new kernel patch
> comes out.  If your systems does the work it's supposed to,
> leave it alone. ;)

ACK.

Quote:>> Unless you have a lot of time to waste , you cannot keep up to
>> date with the kernel.
> Why do you need to?
>> I think that kernels should be released slightly slower ( i am
>> talking about patch versions).
> I think that the sooner the code is out there for people to
> play with, the sooner the bugs get fixed.  If you don't want to
> be part of the bug-fixing process, then don't update every time
> a new patch comes out.

The more people will test, the better will be the "outpointing"
of all kinds of systems. Or how do you want to test any config-
uration?

HAND
Dietrich

 
 
 

Linux kernel development going too fast?

Post by Grant Edwar » Sat, 01 Sep 2001 04:26:10



>> I think that the sooner the code is out there for people to
>> play with, the sooner the bugs get fixed.  If you don't want to
>> be part of the bug-fixing process, then don't update every time
>> a new patch comes out.

> The more people will test, the better will be the "outpointing"
> of all kinds of systems. Or how do you want to test any config-
> uration?

Obviously.  But I can't afford to have some systems involved in
the testing process.

I depend on Linux systems to get my daily work done, and I
can't afford to have those particular systems down. Those
machines don't get upgraded willy-nilly.  

There are other machines that I use for testing and
experimentation. Those get upgraded more often.

--
Grant Edwards                   grante             Yow!  Four thousand
                                  at               different MAGNATES, MOGULS
                               visi.com            & NABOBS are romping in my
                                                   gothic solarium!!

 
 
 

Linux kernel development going too fast?

Post by Dietrich Schaffe » Sat, 01 Sep 2001 06:00:44




>>> I think that the sooner the code is out there for people to
>>> play with, the sooner the bugs get fixed.  If you don't want to
>>> be part of the bug-fixing process, then don't update every time
>>> a new patch comes out.

>> The more people will test, the better will be the "outpointing"
>> of all kinds of systems. Or how do you want to test any config-
>> uration?
> Obviously.  But I can't afford to have some systems involved in
> the testing process.
> I depend on Linux systems to get my daily work done, and I
> can't afford to have those particular systems down. Those
> machines don't get upgraded willy-nilly.  

I agree with you.

Quote:> There are other machines that I use for testing and
> experimentation. Those get upgraded more often.

Therefore I use my home machine, business machines are no-no.

HAND
Dietrich

 
 
 

Linux kernel development going too fast?

Post by Vassilis Pand » Sun, 02 Sep 2001 02:35:51


Ok. Many people dont run the latest kernels. Unless they have a
serious reason not to , shouldn't they be doing that? Whats the point
of developing new kernels , just to develop? Research? Yes, but then
that could take place in a university or something.

And ,you mentioned that kernel releases are often so that bugs can be
caught faster. This is true - to some extent. Normally , "stable"
releases should be the ones without the bugs ( as much as this can be
done; there are always bugs). Shouldn't the unstable versions be
tested for something more than a week?

 
 
 

Linux kernel development going too fast?

Post by el.. » Sun, 02 Sep 2001 03:48:46




>First of all don't get me wrong. I like the way Linux kernels are
>released , and above all , the speed in which security exploits are
>fixed. But , aren't we going a little too fast?

No.

Quote:>It seems to me that
>Linux kernels are released too fast for someone to cope with.

Cope with?  What kind of coping do you think people have
to do?

Quote:>Unless
>you have a lot of time to waste , you cannot keep up to date with the
>kernel.

So what?  Most people don't upgrade kernels all the time.  Why upgrade
for no reason?

Quote:>I think that kernels should be released slightly slower

I doubt the kernel developers will care.

Quote:>( i am
>talking about patch versions).

What's your definition of "patch versions"?

Quote:>This will allow developers to test their
>code a little more ,and before the kernel is deemed "stable" , fix the
>very obvious bugs. Real rapid kernel changes should normally only
>happen when there is a security exploit. Then a new kernel can be
>released ,without having anything more than the previous one; just the
>fixed security exploit. Then stable kernels will be stable , and
>development development.

Do you actually do any kernel development yourself?

--
http://www.spinics.net/linux

 
 
 

Linux kernel development going too fast?

Post by el.. » Sun, 02 Sep 2001 03:55:22




>Ok. Many people dont run the latest kernels. Unless they have a
>serious reason not to , shouldn't they be doing that?

Huh?  Why should anybody change a kernel that is working well unless
they have a "serious reason" to do so?

Quote:>Whats the point
>of developing new kernels , just to develop? Research? Yes, but then
>that could take place in a university or something.

Kernel development certainly isn't scheduled to make *you* happy.  Nobody
needs to justify it to you.

Quote:>And ,you mentioned that kernel releases are often so that bugs can be
>caught faster. This is true - to some extent. Normally , "stable"
>releases should be the ones without the bugs ( as much as this can be
>done; there are always bugs). Shouldn't the unstable versions be
>tested for something more than a week?

It is done how it is done.  If you can't handle that, go run Windoze
and install sircam.  I don't see your name in /usr/src/linux/CREDITS
anywhere so you don't have much credibility when it comes to whining
about release timing.

--
http://www.spinics.net/linux

 
 
 

Linux kernel development going too fast?

Post by Vassilis Pand » Sun, 02 Sep 2001 16:57:21





> >First of all don't get me wrong. I like the way Linux kernels are
> >released , and above all , the speed in which security exploits are
> >fixed. But , aren't we going a little too fast?

> No.

> >It seems to me that
> >Linux kernels are released too fast for someone to cope with.

> Cope with?  What kind of coping do you think people have
> to do?

Change the kernel. Why should people be running an - compared to the
latest
kernel versions which are ( and are supposed to be) stable - unstable
kernel
version , which has bugs already fixed,  and security exploits already
fixed.

Quote:> >Unless
> >you have a lot of time to waste , you cannot keep up to date with the
> >kernel.

> So what?  Most people don't upgrade kernels all the time.  Why upgrade
> for no reason?

> >I think that kernels should be released slightly slower

> I doubt the kernel developers will care.

> >( i am
> >talking about patch versions).

> What's your definition of "patch versions"?

"Pre-release" versions.

Quote:> >This will allow developers to test their
> >code a little more ,and before the kernel is deemed "stable" , fix the
> >very obvious bugs. Real rapid kernel changes should normally only
> >happen when there is a security exploit. Then a new kernel can be
> >released ,without having anything more than the previous one; just the
> >fixed security exploit. Then stable kernels will be stable , and
> >development development.

> Do you actually do any kernel development yourself?

Yes, although i don't really have any experience.

Quote:> >Whats the point
> >of developing new kernels , just to develop? Research? Yes, but then
> >that could take place in a university or something.
>Kernel development certainly isn't scheduled to make *you* happy.
Nobody
>needs to justify it to you.

No , "Just for fun" is a sufficient answer. But i wonder why should
kernel developers develop kernels at that rate , just for a small
percentage of people , because the rest wont upgrade because they dont
have "a serious reason to do so".

Quote:>>And ,you mentioned that kernel releases are often so that bugs can
be
>>caught faster. This is true - to some extent. Normally , "stable"
>>releases should be the ones without the bugs ( as much as this can
be
>>done; there are always bugs). Shouldn't the unstable versions be
>>tested for something more than a week?

>It is done how it is done.  If you can't handle that, go run Windoze
>and install sircam.  I don't see your name in /usr/src/linux/CREDITS
>anywhere so you don't have much credibility when it comes to whining
>about release timing.

Oh , now someone needs to be in /usr/src/linux/CREDITS in order to be
able to express an opinion? Sorry , didnt know that.
 
 
 

Linux kernel development going too fast?

Post by Vassilis Pand » Sun, 02 Sep 2001 22:04:57





> >First of all don't get me wrong. I like the way Linux kernels are
> >released , and above all , the speed in which security exploits are
> >fixed. But , aren't we going a little too fast?

> No.

> >It seems to me that
> >Linux kernels are released too fast for someone to cope with.

> Cope with?  What kind of coping do you think people have
> to do?

> >Unless
> >you have a lot of time to waste , you cannot keep up to date with the
> >kernel.
> So what?  Most people don't upgrade kernels all the time.  Why upgrade
> for no reason?

> >I think that kernels should be released slightly slower

> I doubt the kernel developers will care.

> >( i am
> >talking about patch versions).

> What's your definition of "patch versions"?

> "Pre-release" versions.

update:
Sorry, misunderstanding , which is my fault. When i said patch
versions , i meant the 2.4.n part of the version number. Pre-release
versions SHOULD  be released often , as this would allow better bug
tracking ( this is what im saying from the begining ).
 
 
 

1. 8 Members --- 12 to Go Linux Development

Hello,

We are still looking for 12 more developers to help create and design 3
websites and the software that goes along with it.
Currently there are 6 extremely cool projects in the works.

1. LinuxPorts SSL/GTK
2. LinuxPorts LinCess (A GTK Database similar to M$ Access)
3. LinuxPorts VirtMail (email personally for information on this one)
4. LinuxPorts.Com website
5. LinuxJobs.Com website
6. CMDPROMPT.COM website

We are looking for volunteers.

Poet

2. Difference between xterm and XTerm.

3. Linux Game development goes mainstream

4. news server newsgroup date file

5. Does Linux go fast?

6. X and CL-GD7548

7. Going, going, going, GONE! Bye Windows!

8. hundreds of defunct processes

9. cow goes moo moo, kernel goes oink oink

10. Quiper - Fast Web Application Development Toolkit

11. Fast ADC board or PCI card development tools ?

12. Fast PCI & VLB card for linux which is fast in windows too!

13. linux kernel IDE development process question