The members of X.Org announced the availability for Public Review of the proposed enhancements to the X Window System for the support of IPv6. The proposed code and documentation enhancements for both the X Window X11R6.6 Sample Implementation and the XFree86 V4.3.0. As agreed by the members of X.Org, the review of the proposed IPv6 enhancements will be open until 30th May 2003.
Public Review of IPv6 Proposed Enhancements for X11
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Just wondering but… where is IPv6 used? As far as I can tell, everything still seems to be on IPv4.
It’s sort of like the changing of the magnetic poles…we don’t know when it’s gonna come. Oh yeah, and all of the birds aren’t going to drop out of the sky like in The Core.
Just wondering but… where is IPv6 used? As far as I can tell, everything still seems to be on IPv4.
I agree, I fail to see the reason why the X developers are pay attiention to the IPv6 instead improvement the X. IPv6 isn’t stable and full ready as IPv4; a lot of ISP don’t support it yet.
Now I can finally upgrade to IPv6
just kidding…but honestly, this is a good thing, the more software that supports IPv6 the more likely ISPs and backbone providers will be to upgrade to IPv6.
Just wondering but… where is IPv6 used? As far as I can tell, everything still seems to be on IPv4.
IIRC there are commercial IPv6 providers in Japan and Korea, and I don’t know if there are in China and Taiwan.
Our Asian friends will like knowing that X wants to support IPv6. No wonder most of the IPv6 developers on Linux and *BSD came from Asia.
The specs have already been finalised. Heck, a number of operating system vendors have already bundled it with their latest offering and labelled them as “production ready”. Windows 2003 and Solaris both come to mind.
However, I agree, X does need an overhaul. It is time for the X developers of the world to stop making excuses for poor performance and start correcting the problems. This, however, is compounded by the fact that we have SUN unwilling to make their X server compliant so that Xft and Fontconfig can be used with it. If anyone is holding back X development it is the commercial companies like SUN who are stuck in some sort of time warp.
Yes, I know about the Xft alternative SUN is supporting, however, they’re only supporting it because their X implementation sucks and it would be hell on earth trying to get their X up-to-speed just so that Xft can be used.
Nothing would please me more than SUN being bought out my IBM and some real leadership taken in the Java and *NIX realm.
IPv6 is much more convenient for me, and my machines are already running IPv6, with the machines on a poor ISP link using 6to4 and the other machines using natively provided IPv6 upstream.
On most decent systems (ie not Win95 or BeOS, or that sort of nonsense) it’s just one command line, or one button push to enable IPv6, and of course you can continue to use IPv4 to access most of the common web sites, etc. that remain on IPv4 for the next few years.
IPv6 is a breath of fresh air, back to the Internet as intended, with all my systems being peers, rather than some of them becoming 2nd class citizens behind a tangle of network translation.
Yes, I know about the Xft alternative SUN is supporting, however, they’re only supporting it because their X implementation sucks and it would be hell on earth trying to get their X up-to-speed just so that Xft can be used.
OpenWindows as bundled with Solaris 9 4/03 fully supports Xrender. It works perfectly on the framebuffer on my B2ks (Expert 3D), however I have heard discussion of problems with Xrender support on lower end framebuffers.
stsf is an alternative to xft2, not to xft. In fact, stsf supports the xft API. The Sun mentality is “You did it once and got it wrong. If you’re going to try to do it again, let us try too.”
stsf has technical advantages over xft2, the primary drawback is that it is currently an incomplete implementation, whereas xft2 is more or less complete. Unfortunately, thanks to the xft/xft2 fiasco, applications are in a state of limbo, supporting one or the other of two divergent standards. Development snafus like this are why creating X applications becomes a nightmare… the APIs are all moving targets.
I hate to repeat what others have said, but…
With all the other issues with driver support, fonts, etc they actually want to waste time on this?
Amazing.
I’m confused because I seem to read different information about this all over the place:
Some sources I have seen say that NAT is bad for security, other sources I have seen point out that it is useful.
I can understand the useful argument – basically you aren’t exposing a hold load of IP addresses, only one, right?
IPv6 means NAT isn’t needed any more to conserve addresses, but does NAT still have a role to play, or am I supposed to be better off without it?
Thanks!
I’ve done a little scrounging around, and it looks like there is some debate about whether or not NAT provides any security that could not be provided by other means.
“NAT is NOT necessary for IPv6 security
Kazu, KAME Project, 07/08/98
Many people believe that NAT(network address translator) is necessary for IP security. This is completely wrong.
Recently, many sites install firewall to protect their security. In many cases, NAT is used combined with firewall. This probably lets people misunderstand that NAT is essencial for firewall.
The main purpose of NAT is to save the IP space assigned to site. As you know, the current assign policy of IPv4 is really severe. Many sites can’t get enough IPv4 space. So, they are drived to use private IPv4 addresses.
To communicate a host with a private address in a site and a host with a global address in the Internet, address conversion is necessary. This convertion makes one-way connection. That is, outgoing connections are OK but incoming connections are rejected.
This feature is considered useful to protect site security. But it can be accomplished by other technologies. A good example is filtering. Again, the point is that NAT is not essential for firewall.
By the way, one of the biggest advantages in IPv6 is its huge address space. Each and every site can get enough IPv6 addresses(by default, 16bit subnets, each subnet has 64bit identifiers). So, we don’t have to save IPv6 address space. This means that NAT is completely unnecessary.
To implement one-way connection for IPv6 site, you can make use of filtering. If you want two-way connection environment, just use IPv6 without filtering. You should understand that the two-way connection environment cannot be implemented with IPv4, where NAT is required.
NAT is necessary evil for IPv4. But NAT is evil itself for IPv6. IPv6 is desired to eliminate any and all NATs in the Internet.”
On the other hand:
http://www1.ietf.org/mail-archive/ietf/Current/msg07427.html
Sorry everyone, I should have looked before I asked eh.
>>> OpenWindows as bundled with Solaris 9 4/03 fully supports Xrender. It works perfectly on the framebuffer on my B2ks (Expert 3D), however I have heard discussion of problems with Xrender support on lower end framebuffers. <<<
However, try running Solaris on x86 hardware, it sucks, and boy it sucks BIG TIME. That is where the interest for Solaris is coming from, not their overpriced, underperforming workstations with a smaller selection of software than SGI.
>>> stsf is an alternative to xft2, not to xft. In fact, stsf supports the xft API. The Sun mentality is “You did it once and got it wrong. If you’re going to try to do it again, let us try too.” <<<
How long is it going to take SUN to finally get GNOME 2.2/2.4/2.6 to support it? are they willing to do the heavy lifting? As for the GNOME implementation, it is too little too late. Btw, why does SUN support GNOME yet none of their software use any of the facilities provided by it? where are the GTK+ applications? why does SUN still support SMC which is slow, buggy and pathetic attempt to make Solaris “easy to manage”.
>>> stsf has technical advantages over xft2, the primary drawback is that it is currently an incomplete implementation, whereas xft2 is more or less complete. Unfortunately, thanks to the xft/xft2 fiasco, applications are in a state of limbo, supporting one or the other of two divergent standards. Development snafus like this are why creating X applications becomes a nightmare… the APIs are all moving targets. <<<
They’re moving targets because there is no organisation/people willing to say, “hey, lets get this thing standardised”. There are too many pre-maddonna’s and zealots out there unwilling to compromise during the standardisation process and thus as a result, nothing ever happens.
here are too many pre-maddonna’s and zealots out there
“pre-maddonna’s”????? Either you are a comedian genius or you meant prima donna’s
Just reading the wording of this post reminds me of what I’ve heard about how X is maintained- way too much beurocracy, not near enough action.
Correct me if I’m mislead, though!
If you want to take a crap shot at SUN’s (lack of) interest in X11, consider that SUN once used their own windowing system (NeWS), and their experience with X11 over-running them may have left many traumas.
:-), yes, A big balls up on my part. Thank goodness we have the spelling and anology police on this forum 😉
Well, they were one company against a tide of openstandards.
SUN needs to realise that X is the standard, XFree86 is the defacto standard on *NIX/*BSD whether they like it or not and it is up to them to maintain compatibility with XFree86.
One only needs to look at their terrible directory structure as an example of their shoddy design. There are links from /usr/openwin/lib to /usr/lib and a link from /usr/lib to /lib. What the bloody hell is happening there? are developers that stupid that they can’t use the flag -L/usr/openwin/lib to specify the location of openwin libraries?
Yup, I guessed it. There are lots of people here complaining and whining about things that have nothing at all to do with this news.
For example, Mutiny wondered why “they” are wasting time on IPv6, while having driver, organization, speed issues, which are more important (to who?)…
Just who does Mutiny think “they” are. This news item refers to X.Org, while all these driver etc. issues discussed recently are related to XFree86, which has no say over what X.Org does, and X.Org most certainly is not going to be improving XFree86’s driver support.
I think Mutiny was confused with XWin.org.
If Mutiny did not confuse them, then he shouldn’t have mentioned what he did about drivers,etc.
However, try running Solaris on x86 hardware, it sucks, and boy it sucks BIG TIME. That is where the interest for Solaris is coming from, not their overpriced, underperforming workstations with a smaller selection of software than SGI.
I have never ever in my life seen Solaris on x86 used anywhere. It is most definatly *not* where the interest in Solaris is coming from. Solaris is used pretty much exclusivly on special Sun hardware. And it works very very well on this hardware. Ever try hotswapping your cpu without bringing down a system or even stopping services? You can do it with Solaris on a multiprocessor machine.
SGI is really hurting these days. I wouldn’t use them as an example of a successfull company. Did you know SGI runs Windows on their x86 boxes?
Anyways, STSF’s advantage over xft2 is that it has a far more general abstraction layer. In fact, current implementations of STSF use Freetype as a backend just as xft2 does. This is not a *partial implementation*, it’s how STSF was designed. SUN never intended to recreate the whole font system, just give more flexibility in the rendering engine used.
Please keep in mind that SUN is also not in the Linux ‘desktop race’. Solaris has a very different target audience, and very few people seem to understand that.
One only needs to look at their terrible directory structure as an example of their shoddy design. There are links from /usr/openwin/lib to /usr/lib and a link from /usr/lib to /lib. What the bloody hell is happening there? are developers that stupid that they can’t use the flag -L/usr/openwin/lib to specify the location of openwin libraries?
This is very likely to do with backwards compatibility. A SUN admin can hop on to any SUN box and know where things are. Thanks to all the great choice, distributing a Linux binary that works on all distro’s in nearly impossible, and knowing one Linux distro does not mean you know the layout of another. A certain amount of backwards compatiblity is a good idea when you have a large existing user base that has been around for a long long time. Making a few sym links to get this kind of compatibiliy is a very small price to pay.
Chances are, for every thing you look at as *stupid*, there is a good reason for. Please, before someone goes on to bash someone like SUN, please stop and think for a second. Don’t fool yourself, the engineers at places like SUN are at least as smart as you and me. They don’t do things for no reason. Usually we just don’t know the reason.
Are there alternatives to X?
an example http://www.directfb.org
Not really. Nothing, and I mean nothing, beats X.
The OS/Hardware/Network transparency is just miles ahead of anything else.
X is not bad – ppl just don’t bother to learn how X works.
Sure – X may suck at times, but 90% of that suckage is due
to bad application programmers.
X may suck at times, but 90% of that suckage is due to bad application programmers.
This is suspiciously starting to sound like X sucks so much it’s impossible to make good applications for it
The “X sucks, they should concentrate on making it faster instead of IPv6” is bemusing to say the least. Firstly, X is not much slower than windows if at all, and if correctly set up and with a good driver. Good X drivers don’t exist for every card yet unfortunately, but the results of good drivers are known to everyone using one.
Secondly, you can’t blame X for slowness if you are running gnome or KDE on a 486 with 32M of RAM. There are plenty of nice window managers without the unnecessary fluff. IceWM is reporting a size of less than 4M in top with 2.5M resident. I love it, lots of configurability and keyboard shortcuts and purty to boot. I don’t give a rat’s arse for transparent xterms that go bing!
Thirdly, IPv6 is coming sooner or later and everything will have to be ready for it. In fact it is already here and I could run my home network on it if I wanted. Most of my FreeBSD daemons are IPv6 enabled. So the more years of testing and honing IPv6 on X the better before it becomes the default internet standard.
Lastly, working on IPv6 does not preclude improving drivers and it is probably the case that the people working on Ipv6 are not the same people working on graphical drivers by and large.
Secondly, you can’t blame X for slowness if you are running gnome or KDE on a 486 with 32M of RAM. There are plenty of nice window managers without the unnecessary fluff. IceWM is reporting a size of less than 4M in top with 2.5M resident.
IceWM basically allows you to shift around windows. And it eats 4 MB? And you say that it is light, and without fluff?
SUN Smart? If they were atleast 1/2 as smart as the average person they would have moved their production off to china, contracted their chip production out to UMC and TSMC (TI would lose their contract) and they wouldn’t have a large portion of their software production in the US either. The US is the third most expensive place to do business in, if people in SUN haven’t got that through their heads then god help them.
No IceWM is not the lightest WM but it uses less than a fifth of a freshly started KDE session on the same machine, which could easily double after a while. TWM is using 2.7M on another machine so 4M is not that big. IceWM has features that TWM doesn’t: menu, toolbar. Blackbox, mwm, fvwm, and others are all equally excellent. That’s not my point; why would you need dozens of Megs (KDE, Gnome) to “shift around windows”?
For shifting around windows, even a megabyte is too much. The only minimalistic aspects of those windows managers seem to be in the feature department. They still waste memory like there’s no tomorrow, without any reason AFAICT.
I bought a Gig of SDRAM for about $AU .25c per/M a week ago, I think I can live with 4M of consumption. As for features, they are not that minimal. lwm is minimal. I don’t want minimal in that sense. I want some functionality like keybinding, ability to handle logging out, virtual desktops, menus, etc.. Why IceWM et. al. use 2-4M of RAM, I can’t answer not being a programmer, but it doesn’t seem too excessive to me compared to KDE/Gnome. If you are looking at XFree for embedded projects you may be mad.
“I bought a Gig of SDRAM for about $AU .25c per/M a week ago, …”
… and if an application runs like a dog then all I have to do is keep upgrading my processor every 6 months.
It’s a spurious argument for application bloat when related to processor time, it’s an equally spurious argument when set in terms of memory requirements.
Applications are just stupidly big, and mind bogglingly overloaded with features atm. Pity that trend isn’t going to change any time soon.
“Wow, look, you can play battleships on it.”
“Yeah dumbass, but all I wanted was a text editor.”