* Posts by Vic

5860 publicly visible posts • joined 7 Dec 2007

Google's 'clean' Linux headers: Are they really that dirty?

Vic

Oh dear...

> You can't in general point at the "non-copyrightable portions" of a copyrightable work

Yes, you can.

If the copyright owner disagrees that those portions are non-copyrightable, then you have a claim for breach of copyright - but that does not mean the claim will be successful.

> Shaking off the copyright is much more difficult than starting with something

> that can be said to be exempt

You might be right. But that does not mean that it is impossible to remove the copyrightable portions from copyrighted headers - particularly when those headers specifically permit you to do so.

Vic.

Vic

Jumping the gun...

> The legal question is what that point is and whether Google have crossed it or not.

Not yet it isn't.

That legal question will not arise unless the copyright owners decide to pose it. Not only have they not done so, they have explicitly permitted such behaviour in the distribution of their code.

That means that there is no legal question to answer - and thus talk of what judges might do with this must fall somewhere between "fallacy" and "deliberate scaremongering".

Vic.

Vic

@Gumby...

> I suggest you go back and look at the arguments being made.

Errr - you should do likewise.

> Google is suggesting that they can strip a copyrighted document to a point

> where it loses its copyright

No they are not. This is crucial: they are not doing that. However many times you repeat that claim, it remains factually incorrect.

What they are doing is redistributing the non-copyrightable portions of the original. they are permitted to do this - both by law and by an explicit release within the kernel source.

> Lawyers opposing this notion say that you cannot do this.

Lawyers paid by Microsoft say you cannot do this. All the other lawyers who have weighted in on the subject say you can.

Perhaps there is a correlation there...

> Your reference to the SCO/Linux issue is not a good parallel.

Actually, it is - since the same subject came up, and was shot down for the same reasons.

That SCO were later shown not to have any copyright on those files in the first place was just delicious.

Vic.

Vic

@Gumby

> Unfortunately you're wrong.

Not so.

> Look, here's the underlying issue. Or at least one of them... you cannot

> take software that is copyrighted and 'filter' it to the point where the resulting

> code is *not* *a* *derivative* *work*

It has already been determined (in SCO vs. Novell, amongst many other cases) that you very much can. SCO thought they could prove that Linux was copied from Unix because it had the same numbers in errno.h. This argument went down in flames, because that is not a breach of copyright.

> Essentially they are rewriting the copyright laws.

No they are not.

Copyright laws already permit verbatim copying of de minimis fragments - it would be ludicrous to do otherwise - you couldn't write a book review.

> Notice how Honeycomb isn't open? Now why do you think that is?

Honeycomb is an entirely different matter. It's one of the reasons why I dislike the Apache licence - but it has nothing whatsoever to do with this.

> The chocolate factory *is* *evil*.

It probably is. But once again, you make the mistake of believing that, because you dislike someone, they are obviously guilty of any allegation you throw at them.

I'm not arguing that Google is a fluffy-bunny paragon of virtue. What I'm arguing is that there are no legs to this allegation, by virtue of Google having acted entirely within the way the kernel headers are supposed to be used - as explicitly laid out in the copyright notice of those headers. Can you really not see the difference?

> You're just in denial.

No, you're just wrong. Again.

Please read up a bit on copyright law before you start making ludicrous assertions. You do not understand either US or UK law in this respect.

Vic.

Vic

Poor strawman, that.

> So, if i rip out the forward pages of a book where it states copyright, that

> book is no longer the intellectual property of someone else?

No.

This isn't about removing copyright notices, this is about removing all copyrightable material from the source.

If you took a book and removed all the copyrightable material, you would indeed be able to redistribute whatever is left under your own copyright. But there wouldn't be much left.

Vic.

Vic

95%?

> If Florian Mueller says something, you'll be right 95% of the time if you hold the opposite opinion..

Only 95% of the time?

Vic.

Vic

First things first...

> But this is the USA where how a case is decided

Before a case can be decided, you've got to have a case.

To have a case, you've got to have a plaintiff.

There is no plaintiff. No-one has made a complaint.

This is simply noise from a Microsoft lawyer. The owners of the copyrights involved - the only people who could actually take any action here - are fine about the whole thing.

So - someone whose client would benefit directly from you being fearful, uncertain, or doubtful about using their competitor's product is spreading a little FUD about that product. Now who'da thunk it?

If the kernel owners decide to complain[1], there might be a story here. They haven't yet. That leaves something of an absence of story...

Vic.

[1] And, given that what Google has done seems to be explicitly permitted by the copyright notice on the kernel, I can't see them doing so.

Vic

Re: Linus

> Torvalds is not a lawyer

But he is the owner of a significant amount of the work we're talking about. He has the right not to care, and him saying that he doesn't would constitute a very effective defence should anyone else decide they did.

> doesn't mean that this is not a copyright contravention

Perhaps not - but the COPYING file distributed with every copy of the kernel source says it isn't, and that means it isn't.

That's about as clear as you can get...

Vic.

Vic

Re: Fundamentally Flawed Legal Analysis

> programs that use Linux are not derivative works.

Indeed.

From the COPYING file that ships with every copy of the kernel source :-

<quote>

NOTE! This copyright does *not* cover user programs that use kernel

services by normal system calls - this is merely considered normal use

of the kernel, and does *not* fall under the heading of "derived work".

</quote>

The only unknown here is why the media keep repeating obviously bogus claims from lawyers in the pay of Microsoft, when the copyright owners (such as Torvalds) have already said that there is nothing to see here.

Vic.

Fukushima scaremongers becoming increasingly desperate

Vic

Really?

> could 20-60% be quite reasonably described as 'nearing Chernobyl levels'?

The speed limit on UK motorways is 70mph. 20% of that is 14mph.

Could 14mph be reasonably described as "nearing the national speed limit"?

Even 60% is 42mph. That's not "nearing" the speed limit either.

I don't know whether 20-60% of Chernobyl levels is serious, but it's certainly not "nearing" Chernobyl levels.

Vic.

What it takes to get your desktop back up and running

Vic

Desktop rebuild is easy.

I have kickstart files that describe how the machine should be built, and I have a Cobbler server to run the rollout.

Set the machine in question to be reprovisioned (via the web GUI), and reboot it. A few minutes later, the machine is rebuilt.

If there was anything in /home or /etc that needed salvaging first, it's the work of a few minutes to boot with a USB key and copy data off the HDD before reprovisioning. There's almost certainly a way to do this through a Cobbler setup too, but I've never had to find out how.

Vic.

Google copyright purge leaves Android developers exposed

Vic

I'm pointing...

> 1) Google Bionic is linked against GPL'd kernel headers, therefore it should fall under the GPL.

Linus Torvalds disagrees with you. See http://www.itworld.com/open-source/140916/android-sued-microsoft-not-linux (down towards the bottom of that article). There are other quotations as well.

Richard Stallman also disagrees with you. See http://lkml.indiana.edu/hypermail/linux/kernel/0301.1/0362.html .

> Now point a flaw in that argument.

The flaw is that you assume that use of content from the kernel headers necessarily constitutes a "derivative work" for the purposes of copyright law. This is incorrect; de minimis snippets and essential interoperability copying do not constitute derivation (or else it would be very difficult to use words like "the" and "and", as they are part of copyrighted works).

> Thankyou.

You're very welcome.

Vic.

Vic

RMS...

> Richard Stallman has long since proved that he's willing to pursue legal action

> against GPL violators

And he has also stated categorically that he does not consider this sort of thing to be a GPL violation. See http://lkml.indiana.edu/hypermail/linux/kernel/0301.1/0362.html .

Vic.

Vic

And iterate...

> How is my answer irrelevant when it's correcting a misunderstanding of the OP?

Because it does no such thing.

The poster to whom you replied was telling you why his code is not covered by the GPL. So your attempt to tell him about why the GPL covers derivative works is irrelevant - because his work is not a derivative of the GPL. That's what he was saying...

> Also Google is not using a BSD derived libc,

Well, it is.

Some guy with links to Microsoft has recently claimed that it contains GPL code - I've yet to see any of the copyright owners making similar claims. If any do, then there might be a case to answer[1]. But until and unless that situation occurs, there is nothing to write about here; it's simply FUD.

> The FUD here is because even if you read the licences there's always

> a margin for legal interpretation.

No there is not. The licences are clear.

The complications arise because ACs on the Internet make sweeping statements about what the licences say without actually reading and understanding them.

> But don't let this stop you from making rash statements about what others say.

Pot, kettle, ..

Vic.

[1] Even if there is some verbatim copying from the Linux headers, that does not necessarily mean any violation of copyright has occurred; some things are not protectable by copyright. Exactly that argument bit SCO in the arse when they tried to claim that Linux was a derivative of Unix because it used the same values in errno.h.

Vic

The trouble with claiming someone doesn't know what they're talking about...

> GPL doesn't apply only when compiling in files, linking to them is enough.

True, but irrelevant to the rest of your post.

> a great number of applications link with the Google C library which includes the kernel headers.

If Google were using the GNU libc, the licence would be LGPL, not GPL. You can link against an LGPL library without your code becoming covered by either GPL or LGPL.

However, Google are not using the GNU libc. They're using a BSD-derived libc. That is not GPL...

> So In this case it's better to have some FUD that be FUcked.

I'm not sure I would agree with you there - but it is certainly better not to have to deal with the FUD at all. That's achieved by reading the licences, rather than just making rash statements about them.

Vic.

Government to scrap COI, axe up to 1,000 communication jobs

Vic

Not possible :-(

> The banks should have been allowed to fail.

The banks, in the form they were (and still are) could not be allowed to fail. The interconnectedness of the finance industry meant that no-one would have any money, as any surviving banks wouldn't give anyone else access to anything.

The real problem is that the banks were allowed to get into this situation. If we believe in a capitalist society, no organisation should ever be allowed to get too big to fail, as that places it in a privileged position compared to its competitors, and precludes market forces from correcting errant behaviour.

Quite why the heads of the various banks were permitted to pay themselves large bonuses after screwing the pooch so royally is beyond me. Perhaps their local councils had failed to maintain any lampposts in a condition good enough to support the weight of a fat banker[1]...

Vic.

[1] Yes, I did mis-type that first time around.

Skyhook vows to take Google suit to bitter end

Vic

Read it.

> google IS evil

If Florian says that Google is evil, that's pretty strong evidence that they're saints...

vic.

Wi-Fi security befuddles clueless home users

Vic

People without antivirus...

> they *are* out there and more widespread than anyone thinks.

And I'm one of them[1].

I've been permanently connected to the Internet for many years now, and there have been absolutely no active viruses on either my green or orange LANs. My red LAN has loads - but that's what it's for (it's where customer machines[2] get connected).

So I cannot agree with you that *everyone* should have AV.

Vic.

[1] That's not *strictly* true; my mailserver does have antivirus, because I have many Windows users who connect to it.

[2] Most of those machines *are* running AV, and it's usually pretty much up-to-date - and totally failed, despite lots of congratulatory windows claiming that the machine is fully protected...

Data-mining technique outs authors of anonymous email

Vic

How's that reliable?

In testing, they only had 158 possible authors. they had to fine-tune the algorithm, and they still only got an 80% success rate.

The standard of proof in a criminal case is supposed to be "beyond reasonable doubt". I cannot see how this qualifies...

Vic.

Dim Brits think TARDIS IS REAL

Vic
Thumb Up

The language gets a little bigger...

... As "numberwang" heads for the mainstream :-)

Vic.

Red Hat: 'Yes, we undercut Oracle with hidden Linux patches'

Vic

Re: support

> How much of this is not support then?

It's all support.

> Thinking as an IT customer why do I need support

You don't *need* support. You could decide to use community knowledge on the web and do it yourself - just like you intimated you would with Windows.

But as a commercial entity, that might not make any sense. You might decide it is cheaper and less risky to buy support from a specialist vendor than to try to cover everything in-house.

> (unless I am stupid to throw my money away).

See, it's comments like that that convince me you've never seen a commercial environment.

Users want support. That might be in-house, or it might be bought in. But leaving a bunch of users alone to support themselves is a recipe for disaster except in an environment where all users are sufficiently skilled to be support people if they so chose[1].

So support is a fact of life, whatever OS you're on. I have *far* more Windows support customers than I do Linux support customers, despite the fact that I am targeting the latter.

> Maybe because I don't feel very confident with the software, or there isn't enough

> documentation, either directly from the company or from web forums or I am charitable!.

No. It's because management want to make sure that, should anything tricky come along, the entire company won't be stuck twiddling its thumbs while someone investigates the problem.

A huge number of support contracts are never actually used...

> I would never get support if I know I can fix it myself in the unlikely case something is wrong.

Good for you. But you won't be managing any significant IT resources in the near future. When hundreds of engineers require the IT to work so that they can do anything, every hour of downtime costs you a *lot* of money. Having to trawl through the link-farms on Google to find an obscure fix to a gnarly problem is not a cost-effective way of providing that support function. So you either train up your in-house support staff to be experts in every package they have to see, or you train them to be good enough to deal with the bulk of the workload, and you have a specialist support company on contract to pick up the phone. Guess which of these makes commercial sense once the application count starts to rise...

You'll notice that the above argument doesn't discriminate between OSes, and doesn't mention the quality of the code or its documentation; it's a simple business case that says you mitigate risk by buying support.

> Any explanations why a company would need support if they are looking after their money?

Plenty. You'll see it in action once you start seeing IT in industry.

Vic.

[1] Even then, it will often go horribly wrong because you get competing policies.

Vic

Re: The point of open source

> is to enable someone other than the source code author to compile, debug, maintain the software

Yes. And you can still do that. The source code is published in its entirety.

> subject to legalities such as copyright

Well, that's the real point of Free Software - while the copyright stays with the author, a licence to use and redistribute comes with the code - and such redistribution will leave the recipient with an identical licence to use and redistribute. So al song as you stick to the licence conditions - which is really easy - you can pass the code around to your heart's content.

> the absolute entire point of the GPLs is that your contributions and additions to

> a GPL work are not your exclusive property.

That is incorrect.

Any code you write *is* your property. That's how copyright works.

What the GPL does is to ensure that anyone distributing variants of your code has to pass on the same rights as he got - so others (including the original authors) can get their hands on the source and re-use it.

> They are surrendered to the community

No they are not. This is critical; the GPL does *not* require you to surrender any copyrights.

> Let's have independent security researchers publish information about Red Hat vulnerabilities

Errr - they already do.

> see how Red Hat likes it then

Red Hat would like it very much, I should imagine. They already publish the bugs they know about in public. It's called Bugzilla...

Vic.

Vic

Not relevant

> I seem to remember lots of tut tutting about this sort of behaviour

No you didn't.

> when the aledged chant in the corridors of MS was "DOS isn't done 'til Lotus wont run"

That isn't even close to what is happening here.

Red Hat are *not* breaking any code. They are *not* preventing anything from running on top of the OS. They are *not* trying to influence a customer's choice of applications.

All they are doing is releasing their source code - as much of it as you want - in a single tarball. This is a step down from the previous situation, where they released a "vanilla" tarball from upstream together with a multitude of patches, with full documentation for every one of those patches.

So Red Hat is not quite as good as it used to be - but trying to claim that they are somehow deliberately breaking their code is simply not true.

> Once again we see ideology (Software should be Free!!!) coming up against cold hard reality

> (cash).

And the ideology wins. The software remains Free - as it must do under the licence. All that has changed is that Red Hat is giving away slightly less of the documentation of the source code that it has accrued.

> You have to sell the OS and hope that you get to support it as well, not give it away free but

> hobble it so that others can't offer support.

No. Red Hat demonstrates that to be incorrect - they *are* giving it away and making their money from support. They *are not* hobbling it. You or I could offer support for RH's products, if we chose to do so. Indeed, if we don't change the OS in any meaningful way, it is no harder to do that now than it ever was.

What is a little harder - and, IMO, not so much harder that this will actually make much difference - is to derive something from RHEL that is different in some important fashion. The annotations of which code has been back-ported to older kernels, for example, is no longer there - so we've got to go and look at the code to find out. This is a minor annoyance, not a significant issue.

> This is just ridiculous.

What you're alleging is certainly ridiculous - but then no-one has done it, so I don't really see your problem.

Vic.

Vic

It's not about trust

> You trust them to do that, do you?

They don't have a whole lot of choice. The licence requires them to release their changes under GPL.

Whether they submit any such patches upstream is irrelevant; they *have* to release them downstream (or lose their ability to ship the code at all), and anyone downstream has the right to redistribute under the GPL (e.g. by submitting the patches back to the mainstream development).

The exact definition of "downstream" depends slightly on the method by which source distribution is achieved, but since the code is freely distributable under the GPL, it tends towards being "everyone who wants it".

Vic.

[Waiting for more downvotes to pour in because I've said something that makes FOSS look reasonable again]

Vic

You're reading the wrong lines

> CentOS converts = potential 100B Market cap.

No. CentOS users are not, in the main, candidates for conversion to RHEL. The vast majority of them are well aware of the RHEL offering, and have chosen to take a route that provides them with similar software but no support contract.

So the market cap is closer to being a few tens of thousands than the 100B you guesstimate.

> 20% of current CentOS systems converted to a RHEL Sub =1,600,000

That assumption is your mistake; you dramatically over-estimate the potential.

> Oracle is a problem but not to the same extent of CentOS, IMO

That's because you don't understand Free Software. RH are not selling code, they're selling support. CentOS is not their competitor - it is what gets people used to using RH-style systems. It is a feeder. The existence of CentOS - and other rebuilds - generates revenue for Red Hat, rather than taking it away.

Red Hat know this. That's why RH employees regularly help out anyone who is trying to rebuild their code. RH effectively sponsors all such rebuilds to a small extent. They appear happy to do so - and they are making a vast amount of money whilst following that policy.

> This decision creates huge upside for Red Hat shareholders

This decision makes no difference to RH shareholders. It's a small tweak to the way they ship code. It won't affect CentOS or any other clone-type rebuilders. It will only affect people who want to build something based on RHEL, but differing in some important fashion - which is the sort of thing I tend to do. It makes more work for me, but RH's shareholders have no reason to care about that - it really won't affect them in any way.

> but goes against the spirit of the GPL....

No, it doesn't.

RH are releasing all their source just as the GPL requires them to. What they are *not* releasing is all the annotation they've built up in their VC repository. That is unfortunate from my perspective - and I hope they change their minds when they see this having no real impact on Oracle - but it's absolutely fine as far as the GPL goes. The GPL entitles users to the source used to build their binaries, not to every thought that has gone through the developers' heads.

Vic.

Vic

Re: eh?

> I didn't say impossible. I said "not easy"

But it is easy. You just distribute under section 3(a), rather than 3(b).

Somewhat pointless, but not at all difficult.

Vic.

Vic

What are you talking about?

> Windows Servers, they work and they work very well

If that were true, my bank manager would be very upset.

Windows servers need much more coddling. Linux Just Works.

> Red had and the rest of the Linux crowd are trying to make money by dishonesty

That is entirely untrue. They are extremely honest.

> by making something wrong on purpose so you need their support.

This is not happening. No-one is making anything "wrong on purpose". You appear to have read a different article to everyone else.

What RH are doing is ceasing to lay out in enormous detail every single patch to the sources that they make, rolling them up into is single source tarball instead. If you think you'll get something like that out of Microsoft, you're in cloud cuckoo land.

The difficulty is that RH are now offering less than they previously did, not that what they're offering now isn't still rather wonderful.

> I never had to ask for any support with Windows software from Microsoft directly

Big deal. If you don't want support - don't buy it. But others *do* want it - and the market is there for exactly that. It's not a compulsory charge, it's an offer.

> Why do people need support directly from Redhat

They don't. But if they want it, it's there.

> I can only think there is something wrong with their software.

You have clearly never worked in a commercial environment. How are the GCSEs going?

Vic.

Vic

Please read the article...

> If you aren't releasing source code in useable form, then you aren't releasing source code.

They *are* releasing source in a usable form.

What they are no longer doing is breaking it down into a bazillion tiny patches, with explanation for each and every one of them.

But I suspect someone will.

Vic.

Vic

I smell troll...

> Now that Red Hat has decided to covertly introduce flaws in their system

It has done no such thing.

This change merely wraps up all the individual patches into one super-patch, and they no longer put all the info into the changelog. It's the same code as it ever was - just without all the explanation RH have historically put into their releases.

It's a shame RH has done this, but I can see their point.

Vic.

Vic

This is a pain

RHEL distributions are long-lived, and the kernel shape is the same at the end of life as it was at launch - so anything RHEL4-based started with a kernel that looked like 2.6.9, and still does to this day. But they are very far from obsolete - RH back-ports much new kernel development into these old frames to give something that is updated and fresh, whilst maintaining compatibility with what went before. That's a Good Thing(tm).

However, if you want to find out if a certain kernel incorporates a certain feature, you have to go through the patchset. The availability of the changelog, with patch documentation, is a boon here. Red Hat's removing it will be a royal pain in the arse to those of us who use their code.

I can't blame them. They are operating well within the GPL, and give what Oracle (in particular) has been doing, I can't say I'd have done anything differently. But it is properly annoying...

Vic.

Sony wins subpoenas revealing visitors to PS3 jailbreaker site

Vic

Not yet proven...

> what that law requires us not to do;

Remember that it is only SCEA's *allegation* that any law requires us not to do this; no court has yet upheld that view (nor rejected it).

So SCEA has been granted the right to find out the IP addresses (and thence, according to their argument, the identities) of anyone who has associated with Hotz via his website, and heard any of his ideas, which might well qualify as Free Speech in the US.

I find this ruling rather troubling...

Vic.

Electric cars not as 'green' as advertised

Vic

Found it.

> There was an excellent compromise around a few years back

It is called the Monotracer - see www.monotracer.ch (N.B.: site mostly in German)

And they've got an electric version as well. Stats look really rather good.

I reckon this sort of thing is the future of travel - excellent performance, with excellent economy/emissions, and not much of a compromise for most people.

> Only trouble was that it was *stupidly* expensive :-(

And still is. They *start* at 50 grand. But you'll want options.

Vic.

Vic

There are many alternatives...

> Get a full license, get a low capacity motorcycle/scoot.

Even a high-capacity bike, with performance to put a supercar to shame, will generally give fuel economy a damn sight better than most cars, and with a concomitant lack of emissions.

But bikes aren't for everyone. You do tend to get cold and wet in the winter, and carrying luggage can be a bit of a problem[1].

There was an excellent compromise around a few years back - they'd started with a K1, IIRC, and stretched the frame to give two full seats line astern. Then the whole thing was encased in a fibreglass body. So you've got a vehicle half the width of a car, capable of carrying two people (or one with luggage), with good performance and a small (for a car) engine.

Only trouble was that it was *stupidly* expensive :-(

Vic.

[1] I had an interesting ride back from Bristol once, with an ATX PC on the back. I'm not doing that again...

Vic

Cars with generators in

> Why haven't we seen any cars being developed along those lines

We have. Porsche probably made the first of them - in 1905.

Vic.

Antennagate Redux: Consumer Reports condemns Verizon iPhone 4

Vic

Not just a "me too" offering...

> the HW manufacturers are reduced to "me too" offerings.

No, they're not. They're *offered* the opportunity to be such, whilst simultaneously being *offered* the opportunity to do whatever they like with Android. They don't even need Google's permission to take the Android code and do something different.

> Android strips away their ability to take the initiative

Not at all. Anyone can take any initiative; the fact that so many manufacturers currently are not doing so is because they see value in commonality, not because it is forced upon them.

This situation might change. Or it might not.

Vic.

BT Home Hub 3 ADSL Wi-Fi router

Vic

Bet it's still shit at VoIP

The earlier models of HH do nasty things with SIP packets. I'm sure that's just an accidental error in some sort of SIP ALG, and not a TelCo trying to protect its core business.

What's the betting this box is just as wank?

Vic.

US scientists build laser-killing device

Vic

Isn't it obvious?

They've developed this material especially for coating the targets used in the ALTB tests...

There. That should justify a few more $trillion.

Vic.

SCO: 'Someone wants to buy our software biz!'

Vic

Formerly Glorious

> Sorry, not sure it was ever a "glorious" compagny ...

Oh, it was.

SCO Unix used to be a sound product. And SCO paid Red Hat to put SCO Linux together, which is how they got sufficient funding to launch Red Hat Linux. In days of yore, SCO did plenty of really good stuff.

Then along came Darl.

Vic.

Oz road safety strategy moots mobile phone ban

Vic

Computer says no...

> they are all just as bad, if not worse than holding a handset to your ear.

Wikipedia[1] has a list of some trials that demonstrate that using a phone whilst driving is more hazardous than you might think - see http://en.wikipedia.org/wiki/Mobile_phones_and_driving_safety

I'm somewhat sceptical of one of the studies though, on account of the sentence "It should be noted that the data of this study was adjusted to reflect socially accepted results"...

Vic.

[1] No, I'm not citing Wikipedia as an authoritative source, just a list of studies that might be...

Vic

Re: that old arguement...

> At a junction, a passenger will oftern stop talking when you are trying to get across.

You've not met my missus, then?

Vic.

Stroppy Belgian students in Ryanair mutiny

Vic

Gratuitous YouTube link...

If you've ever flown RyanAir, you'll enjoy this :-)

http://www.youtube.com/watch?v=ZAg0lUYHHFc

Vic.

World shrugs as IPv4 addresses finally exhausted

Vic

@Mark

> couldn't you have a router holding the external IPv6 address and NAT the traffic

The IPv6 spec doesn't allow for this - NAT doesn't exist in IPv6.

But it wouldn't surprise me to see that change as people come to terms with the ramifications of globally-routable devices.

Vic.

Vic

Re: IPv4 and IPv6

> There is no transition, there is no 00:00:ip:v4:ad:dr.

Yes there is; it's the 6to4 gateway system. IPv6 address in the 2002::/16 range are reserved for just this use.

So, if your IPv4 address were 212.100.234.54, you could also use the IPv6 address 2002:D464:EA36::/48.

Alternatively, a dual-stack implementation can encode IPv4 addresses directly as IPv6 addresses - so 212.100.234.54 can be encoded as ::ffff:212.100.234.54 (note dotted-decimal notation of the last 4 octets, just to make things nice and clear... )

Vic.

Vic

Too much momentum

> Just think, if DNS told you what port to use

What, like with a SRV record?

Vic.

Vic

Hmmm...

> How does someone with ONLY IP6 address IP4?

Typically through a 6to4 gateway, which has a prefix 2002::/16.

> They need to rethink IP6 interworking and security.

No. Both internetworking and security are just fine in IPv6. But a lot of people will need to learn quite a bit more about networking when IPv6 becomes prevalent - lots of people are suddenly going to be globally routable.

It would not surprise me to find IPv6 router manufacturers incorporating something akin to NAT so that internal networks can use RFC4193 addresses only, But AFAIK, that's not in the standard at present...

Vic.

Vic

Because...

> If we're used to 192.168.0.X, surely banging a few extra zeros in the middle won't hurt?

You don't exactly get that choice.

The first address you'll get is a "link-local" address, which doesn't leave your LAN. It will look something like "fe80::204:23ff:fe65:9818". Quite a lot of that is defined in the spec, so you don't get too much of a choice about that.

If you actually want to talk to other hosts on the Internet, you need another address - your link-local address is really only useful for acquiring your global address. The top bits of your global address are defined by your provider - you'll never get less than a /64, meaning you've got 2^64 addresses to allocate pretty much however you want, but you'll need to allocate a unique global address to each machine that wants IPv6 Internet access.

Forget about having "private" LAN addresses and NATting at your router - NAT doesn't exist in IPv6-land. Every host that talks to the Internet is going to be globally routable[1]. So you need to do proper firewalling, not just rely on hiding behind NAT.

There's complexity there, for sure, but it will be worth it in the end. But ISPs really need to get their shit together and start supporting IPv6. Until they do, you either ignore IPv6, or you create a 6in4 tunnel. Whilst not especially difficult, that does need some technical nous.

My tunnel provider gave me a /48. That's over 10^24 addresses, and I have no idea how to use that many. Even if I split it down into /64s (minimum allocation unit) and passed then on, I'd still have 64k /64s. I don't know *that* many people...

Vic.

[1] I am deliberately ignoring RFC1884 site-local address scheme, which is deprecated, and RFC4193 private addresses, which cannot be used for communications with globally-routed hosts.

Vic

IPv4 addresses not yet exhausted...

7 /8s have just been released for allocation. That's 117 million addresses that are now available for release to users that weren't available a couple of days ago.

So although things are obviously getting close, it is incorrect to say that the address pool is exhausted.

And it would all last quite a lot longer if people would get the addresses they need, rather than just applying for a subnet because they can...

Vic.

Microsoft discovers disposable email

Vic

Three accounts...

> using more than one account is no harder than using just one.

I have >3500 addresses. A new one for every time I make a new contact.

/etc/aliases is your friend... :-)

Vic.

US lawmakers eye internet 'kill switch'

Vic

You'd need to work out where the rot is first...

> Anyone who writes internet with a capital "I" is obviously a moron

Anyone who writes Internet with a capital "I" obviously knows the difference between "the Internet" and "an internet".

And that's the thing with this idiotic proposal - it really isn't difficult to build your own internet, it's just difficult to get access to everyone else's networks. But there's no reason why any particular network shouldn't simultaneously peer with more than one compatible internet...

Vic.

Openistas question UK.gov's £300k crime-mapping website

Vic

Balance where due...

> IBM "on demand" and extra bandwidth on demand for the first few days

To be fair, they have put the site up on Amazon EC2. That *should* have given them the bandwidth they needed.

Unfortunately, the backend DB seems to be hosted somewhere with insufficient grunt; the 503 messages I get[1] from the front end claim that it is a backend issue.

So it would appear that the webby front end has loads of power and bandwidth, but someone is running the backend somewhere where the best connectivity is a bit of wet string, or running it on something with the throughput of a constipated hamster. Or maybe both.

Vic.

[1] wget is your friend :-)