ipv6 and stock Arch kernel26

pixel · 2006-04-06 11:20:11

not really. I said only one of the minor reasons is apparently connected to Firefox but that's it, and i also explicitly written you can disable it in config:about. How many times i need to repeat it before you get it?

iphitus wrote:

It being there, does not force anybody to use it.

You can't disable it, so that's forcing people to use it. I ask, what's so hard in making ipv6 modular? it's one switch in config. It still can be enabled by deafult if devs really want it, but this time it would be easy to disable it... think about it. I'm not asking to erase it completely from the kernel, what im askin for is only to make it more flexible, because right now you CANT disable it...

Mikos · 2006-04-06 14:21:22

As I said in bugzilla, I am strongly against modularizing IPv6 support. IPv6 is the future and it should be promoted, not given away from kernel as module. IPv6 is standard and should be always in kernel.

And if there is bug in Firefox, then you should talk to Firefox developers, not Arch Linux developers.

Neuro · 2006-04-06 14:32:15

pixel wrote:

I'm not asking to erase it completely from the kernel, what im askin for is only to make it more flexible, because right now you CANT disable it...

I agree. Just tell me, would it hurt anyone to turn it into a module? Isn't Arch about simplicity or choice? ipv6 should stay in the kernel, but as a module, so that using it would be a conscious decision in your /etc/rc.conf.

If you want ipv6, that's cool. Load the module, but please: let others decide (and please, no "compile your own kernel BS"). Instead of just just repeating your pointless argument that by keeping ipv6 built-in you allow it to gain momentum, do something could actually benefit the new protocol. Write a wiki entry that would tell users how to configure the tunneling in Arch, describing what are the pros (and cons) of the new protocol, showing where to find ipv6 gateways, etc.

PJ · 2006-04-06 14:45:44

Wouldn't it be possible to make it in to a module and to have it loaded with initrd/initramfs. This would make it possible to have an option in the mkinitrd.conf/mkinitramfs.conf which default would be set to include it in the kernel.

Mikos · 2006-04-06 14:46:28

No. If you want to make IPv6 as module, then alse make IPv4 as module. IPv4 shouldn't be privileged!

Neuro · 2006-04-06 15:16:24

Mikos wrote:

No. If you want to make IPv6 as module, then alse make IPv4 as module. IPv4 shouldn't be privileged!

There is one, simple reason why ipv4 should be privileged: it's broadly used. ipv6, on the other hand, has marginal use. Even smaller than ppp, and ppp as a whole is a module.

<irony>I can't stand this injustice! ipv4 and ipv6 are built into the kernel! All network protocols should be treated equally! I demand that ppp, atm, (whatever else) be compiled into the kernel!. IPv4 (and IPv6) shouldn't be privileged!</irony>

Mikos · 2006-04-06 16:34:42

Your example is simply wrong. IPv6 is intended to replace IPv4. It is IP done well and it will completely replace IPv4 in future.

Look for example to NFS support in Arch Linux kernel. There is also NFSv4 in kernel, which is supposed to replace NFSv3. This is similar situation like IPv4 vs. IPv6.

codemac · 2006-04-06 16:37:41

The only bad part about having ipv6 built into the kernel is that some applications break when using it.

Ok.

So what does this have to do with our kernel again? Go add comments to their bugzilla, not waste forum space.

It seems odd that users who want a bleeding edge distro are so against ipv6. You type "pacman -Syu" because you want updates. You want the latest and greatest. If you don't, then arch is not the distro for you. Arch can't be everything, and the last thing it will be is disabling new and upcoming techs.

pixel · 2006-04-06 17:25:31

codemac, it seems 76% percent of the Arch community is unfortunately not sharing your view. Maybe it's time for YOU to change your distibution to more "bleeding edge" distro.

It's funny how silly this discussion can get.. Noone really wants to get rid of ipv6. We just want choice, modularized ipv6 is the way to please both camps.
Do you also force people to listen to your favorite music, wear only your favorite clothes etc?

Mikos · 2006-04-06 17:32:13

pixel: It's developers who are creating this distribution and who are in charge what will be and what won't be in kernel, not users. Arch Linux is and (I hope) always will be bleeding edge distribution.

This thread is nonsense flaming...

Mikos · 2006-04-06 17:33:58

And btw. 42 people aren't in any case whole Arch community ;-)

Mikos · 2006-04-06 17:38:56

As I said before: If you want to make IPv6 as module, then alse make IPv4 as module. IPv4 shouldn't be privileged.

Neuro · 2006-04-06 17:41:31

Mikos wrote:

Your example is simply wrong. IPv6 is intended to replace IPv4. It is IP done well and it will completely replace IPv4 in future.

And if it one day does, I'd be happy. But... it's rare at best. Do you know any ISP that supplies connection via ipv6? I don't care about the ipv6 through ipv4 argument, it's only a nice hack, not valid here. It's still "the future", and, unfortunately, a fairly distant one IMHO.

Look for example to NFS support in Arch Linux kernel. There is also NFSv4 in kernel, which is supposed to replace NFSv3. This is similar situation like IPv4 vs. IPv6.

No this is not similar at all. You can't compare network filesystem serving solution with the actual network protocols. It just doesn't work. Moving to NFSv4 is magnitudes easier than moving your internal network to ipv6. You only have to worry about the things you use internally, no need to care about address translation, whole lot of network hardware not supporting the protocol etc. With NFS you only need to upgrade the server and client equipment. And yes, you can say that NFSv4 is the future, but it's a much much closer one than ipv6.

codemac wrote:

It seems odd that users who want a bleeding edge distro are so against ipv6. You type "pacman -Syu" because you want updates. You want the latest and greatest. If you don't, then arch is not the distro for you. Arch can't be everything, and the last thing it will be is disabling new and upcoming techs.

I'm not against bleeding edgeness of Arch. It's great. But running the latest and greatest hardware on your computer is totally different. What your ISP/uni network is running as their network protocol is not your choice. It's not something you can switch like desktop environments or upgrade to the latest version with pacman -Syu. If they're running ipv4, there isn't much you can do about it. Having ipv6 compiled into your kernel statically as a statement to the world "I support new technology", won't help either.

And no, I'm not against using the latest and greatest (I really hope suggesting that wasn't your intention). I really would love to use ipv6 full scale, but I can't. And it ain't my fault. So currently, ipv6 is useless to most of Arch's users. To all you "we must have it built in" guys, one straight question: Why couldn't you live with ipv6 compiled as a module so that other people could (it would be on by default) turn it off?

phrakture · 2006-04-06 17:45:55

pixel, you are aware this is the exact reason ABS is provided for you.

You want a different stock config? Go ahead and recompile it - it's trivial. If you are unhappy about a given decision, it's best not to stomp your feet about it. Write a bug report (you did) and see what happens. If it doesn't change, then just recompile it yourself.

I don't see why this is a big deal. Can you provide exact instances and/or performance numbers where IP6 is a problem? The only problem I have ever heard of had to do with firefox's DNS lookups or something dumb which can be handled by a firefox setting.

Neuro · 2006-04-06 18:07:04

phrakture wrote:

pixel, you are aware this is the exact reason ABS is provided for you.
You want a different stock config? Go ahead and recompile it - it's trivial. If you are unhappy about a given decision, it's best not to stomp your feet about it. Write a bug report (you did) and see what happens. If it doesn't change, then just recompile it yourself.

That's exactly what I did. The only reason why I'm continuing this discussion is that I found some of the statements in this thread intriguing.. to say the least.

I don't see why this is a big deal. Can you provide exact instances and/or performance numbers where IP6 is a problem?

ppp doesn't cause any problems stability or performance problems and it's by far more widely used than ipv6. and still is a module (I know I'm pulling you on this one )

brain0 · 2006-04-06 23:49:44

Mikos wrote:

No. If you want to make IPv6 as module, then alse make IPv4 as module. IPv4 shouldn't be privileged!

ipv4 can not be compiled as a module, at least not without patching the kernel.

pixel · 2006-04-07 00:48:22

phrakture wrote:

pixel, you are aware this is the exact reason ABS is provided for you.

Actually i like compiling my custom kernel without ABS at all, but it has nothing to do with the current discussion. The stock Arch Linux kernel package should be as flexible as possible - i can't understand arguments of people wishing to have ipv4 as a module too.. I dont quite get it if they are really so ironic or just making fools out of themselves. For me it's only logical that ipv6 should be modularized because its usage is still very rare, so why force people to use it?

codemac · 2006-04-07 01:27:07

I still have yet to see an answer to phrakture's question.

List some examples outside of Firefox DNS where ipv6 is at fault.

cactus · 2006-04-07 04:11:05

not just firefox dns, but dns in general. Not many quad A records out there.
dig -t AAAA google.com
dig -t AAAA slashdot.org

I tried a few off the top of my head. About the only one I found having a quad A record was one of the tld servers.
dig -t AAAA a.gtld-servers.net

heh. kinda cool to actually find one though.

something else interesting..
http://www.ripe.net/rs/ipv6/stats/

You can drill down to see who has been allocated ipv6 blocks. heh. Looks like cisco got a ipv6/32 o.O

heh..
Oregon ipv6
cool! go oregon!
ipv6 for trees!

edit: hmm.. comcast has a /32 as well..interesting.
*chin scratching*

mcubed · 2006-04-07 07:34:08

codemac wrote:

If you don't, then arch is not the distro for you. Arch can't be everything, and the last thing it will be is disabling new and upcoming techs.

Perhaps your right, if the Arch Linux attitude is don't listen to what users' suggest. But I don't see how modularizing IPv6 is "disabling" it. It merely makes it easier for those who don't want it to avoid it. Those who want it can enable it without much difficulty. It seems more people, at this stage, don't want it -- and why would they? As currently implemented, it's an ugly hack that causes problems. Here's this "new and upcoming" technology that may or may not actually be going anywhere and seems to be losing momentum, that most people can't make any use of because their ISPs don't support it, and that can and has broken any number of applications (web-browsers, p2p apps, ftp, etc.) that millions of people use everyday. And the choices are:

1) Force it on every Arch Linux user by default, even though most of them can't do anything with it yet;
2) Let those (few) Arch Linux users who might actually be able to make use of it enable it.

And, somehow, you think #1 is the right choice?

iphitus · 2006-04-07 10:30:22

mcubed wrote:

codemac wrote:
If you don't, then arch is not the distro for you. Arch can't be everything, and the last thing it will be is disabling new and upcoming techs.
Perhaps your right, if the Arch Linux attitude is don't listen to what users' suggest.

No, the developers listen to everything the users say. Many developers dont read these forums, so chances are, what you say here, won't be heard. But the bug tracker is followed. And there has been a much more reasonable and more civilised debate about this issue there.

<snip>
1) Force it on every Arch Linux user by default, even though most of them can't do anything with it yet;
2) Let those (few) Arch Linux users who might actually be able to make use of it enable it.
And, somehow, you think #1 is the right choice?

There are, believe it or not, reasons why it is included. First off, if you read phrakture's comments here: http://bugs.archlinux.org/task/4357, as an example, X11 spits a range of errors when it isnt enabled.

I have yet to honestly see a convincing reason why, keeping it is a bad thing. It doesnt harm you when it's installed and in use, and indeed, on most distro's it's transparent and you wouldnt notice.

The only time it's noticable and causes problems, is when an application craps up about it. But unfortunately, that's a fault in that application, and ought to be reported to the respective bug tracker.

And thirdly, some users actually do use it. Mikos has been prominent here, phrakture has stated that he has used it, and thats just two of what is likely to be more than you think.

Please, give me a concisely, level headed reason why having it enabled, disables, effects, or does bad stuff to you, and that reason is not due to a problems in an application's implementation/usage, and we will consider it.

I have yet to see a reason to change it from what it is now. It works, it's fine. Anything else is just personal preference.

Too many posts here have been inflammatory, emotionally or ego driven. Calm down guys. Look at how minor the issue really is.

James

cactus · 2006-04-07 15:51:37

Sorry to be dense iphitus, but one of the reasons you recommend keeping it..is due to an application crapping out when it is not present (xorg).
I am sitting here scratching my head and trying to see the other side. It seems both sides are saying "if it isn't done my way, it causes problems for some applications. Tell me why it should change from my way of thinking".

I prefer it modularized so that it can easily be enabled or disabled. I prefer to have most things additive based. Start with a reasonable default, and add the extras that I want. I have bumped heads with a few people over what I consider "reasonable defaults", and it seems lately that the arch community is migrating towards including certain defaults that I do not really prefer.

I undertand why. Many of these new defaults are coverage configurations. They are useful for the greater populace, but that doesn't stop them from being a pain in *my* butt. The more customization work required on my end to get things "where I want them" makes kitty cry.

This really isn't the fault of arch. Call it a subtle shift in the target audience...Call it an attempt to push the leading edge...Call it whatever you want. In the end, it is just *me* deciding that a default configuration does not suite *me*.

More generallly, I really like this community...most of the time. That has been one reason why I have stuck around even after some of the technical choices don't agree with me. I am powerless to change, so I just stick around and try to help new people..while personally, I am using arch less and less each day (unfortunately).

But discussions like these...I just don't get the vitriol that both sides seem to sling. It seems like ipv6 module inclusion has become some type of religion. Almost like people have ceased thinking, and prefer to beat others over the head with their opinions. I guess this is just something that happens in Linux communities. I just thought the Arch community was a little better.....

meh...maybe I just need more coffee today..stupid friday..

pixel · 2006-04-08 16:07:35

tpowa and devs, thank you. I think both sides can do with ipv6 now what they want in the new kernel26-2.6.16.2-1.

mcubed · 2006-04-10 03:09:34

iphitus wrote:

There are, believe it or not, reasons why it is included. First off, if you read phrakture's comments here: http://bugs.archlinux.org/task/4357, as an example, X11 spits a range of errors when it isnt enabled.

Why would X11 depend upon IPv6? It doesn't look like, from what Thomas posted in the thread you linked to, that those errors are a problem anyway.

iphitus wrote:

I have yet to honestly see a convincing reason why, keeping it is a bad thing. It doesnt harm you when it's installed and in use, and indeed, on most distro's it's transparent and you wouldnt notice.

If that were the case, I wouldn't care in the slightest. It has not been my experience.

iphitus wrote:

The only time it's noticable and causes problems, is when an application craps up about it. But unfortunately, that's a fault in that application, and ought to be reported to the respective bug tracker.

What bug would that be? Why is it that Firefox works without issue Windows and OS X? I wouldn't know how to report a bug that isn't reproducible in that way. I could launch Firefox under Debian and watch it stall out trying to connect to a website, scoot over to my iMac three feet away, launch Firefox and watch it connect to the same website without a problem. Could you explain to me how this is a Firefox bug, because I really don't understand? And the other protocols that weren't working for me under Debian, but worked fine under OS X -- bittorrent, ftp, other IPv6-enabled http browsers like w3m -- those apps are buggy, too? Even though they work fine under OS X?

iphitus wrote:

Please, give me a concisely, level headed reason why having it enabled, disables, effects, or does bad stuff to you, and that reason is not due to a problems in an application's implementation/usage, and we will consider it.

I don't know how to be more concise: I named three internet protocols that were hosed under Debian while working perfectly under OS X. What information do you want?

I really don't care what Arch does as long as it doesn't create the same kinds of problems I was dealing with for more than three months. I am (or was) in the process of switching distros because Arch has been working for me and Debian has not, due to IPv6 issues. If Arch breaks too, then I'll find something else, simple as that. I always have OS X to fall back on. I'm not falling on any sword to try to make IPv6 be modularized. But this thread asked for opinions about IPv6, and mine is negative. If it had screwed things up for you as much as it has for me, yours would be too.

Mikos · 2006-04-10 12:22:14

mcubed: This is really _NOT_ a Linux problem. And it is not problem with IPv6. It is problem with bad application support for IPv6. So yes, you should tell Firefox developers about this problem, not Arch Linux developers or Debian developers. Firefox have bad implemented IPv6 support and should be fixed, it's that simple (this is why it works for you without problems if IPv6 is disabled).

Sometimes it is also problem with misconfigured DNS servers. Again, this is _NOT_ Linux or IPv6 problem, this is problem of administrators of that bad DNS server.

But to tell you truth, I have never had such a problem. So this is not a general problem. And btw. I am actively using IPv6 for more than year without problems (but I haven't got problems also before my IPv6 connectivity, so as I said, this is not a general problem).

Arch Linux

#26 2006-04-06 11:20:11

Re: ipv6 and stock Arch kernel26

#27 2006-04-06 14:21:22

Re: ipv6 and stock Arch kernel26

#28 2006-04-06 14:32:15

Re: ipv6 and stock Arch kernel26

#29 2006-04-06 14:45:44

Re: ipv6 and stock Arch kernel26

#30 2006-04-06 14:46:28

Re: ipv6 and stock Arch kernel26

#31 2006-04-06 15:16:24

Re: ipv6 and stock Arch kernel26

#32 2006-04-06 16:34:42

Re: ipv6 and stock Arch kernel26

#33 2006-04-06 16:37:41

Re: ipv6 and stock Arch kernel26

#34 2006-04-06 17:25:31

Re: ipv6 and stock Arch kernel26

#35 2006-04-06 17:32:13

Re: ipv6 and stock Arch kernel26

#36 2006-04-06 17:33:58

Re: ipv6 and stock Arch kernel26

#37 2006-04-06 17:38:56

Re: ipv6 and stock Arch kernel26

#38 2006-04-06 17:41:31

Re: ipv6 and stock Arch kernel26

#39 2006-04-06 17:45:55

Re: ipv6 and stock Arch kernel26

#40 2006-04-06 18:07:04

Re: ipv6 and stock Arch kernel26

#41 2006-04-06 23:49:44

Re: ipv6 and stock Arch kernel26

#42 2006-04-07 00:48:22

Re: ipv6 and stock Arch kernel26

#43 2006-04-07 01:27:07

Re: ipv6 and stock Arch kernel26

#44 2006-04-07 04:11:05

Re: ipv6 and stock Arch kernel26

#45 2006-04-07 07:34:08

Re: ipv6 and stock Arch kernel26

#46 2006-04-07 10:30:22

Re: ipv6 and stock Arch kernel26

#47 2006-04-07 15:51:37

Re: ipv6 and stock Arch kernel26

#48 2006-04-08 16:07:35

Re: ipv6 and stock Arch kernel26

#49 2006-04-10 03:09:34

Re: ipv6 and stock Arch kernel26

#50 2006-04-10 12:22:14

Re: ipv6 and stock Arch kernel26

Board footer