log in | register | forums
Show:
Go:
Forums
Username:

Password:

User accounts
Register new account
Forgot password
Forum stats
List of members
Search the forums

Advanced search
Recent discussions
- General discussion at Rougol May meeting (News:)
- Wakefield Super Prize Draw (News:)
- Certificate error (Gen:2)
- Wakefield-in-Bradford Show 2022 takes place this month (News:1)
- AMCS free versions are live! (Gen:29)
- Archive gets a new edition and a new website (News:)
- RISC OS Developments releases their new TCP/IP stack (News:5)
- April 2022 News Summary (News:)
- March 2022 News Summary (News:2)
- Rougol April meeting report (News:)
Related articles
- Rounding Up February
- Iconbar in update shocker!
- Iyonix Linux, USB Printer Port, Hydra update
- Catch Up
- Castle Technology deny GPL breach
- Podcast IV - coming soon!
- BBC Domesday online
- April Fools!
- Select 3i4 released [updated]
- RISCOS Ltd. laugh off Castle allegations with Select3i4
Latest postings RSS Feeds
RSS 2.0 | 1.0 | 0.9
Atom 0.3
Misc RDF | CDF
Site Search
 
Article archives
The Icon Bar: News and features: Castle break GPL?
 

Castle break GPL?

Posted by Andrew Poole on 00:17, 8/2/2003 | , , , ,
 
Daniel Barron contacted The Icon Bar this morning to point us at a Slashdot article which suggests that Castle Technology may be breaking the General Public Licence in their RISC OS 5 machines.

The Slashdot article points us to an article on Linux Kernel Mailinglist, which outlines that Castle have "taken some of the Linux kernel 2.5 code, and incorporated it into their own product, "RISC OS", which is distributed in binary ROM form built into machines they sell. This code is linked with other proprietary code."

The article, by Russell King, states that "Having discussed this with Linus, Linus is of the opinion that a public letter should be written to Castle Technology Ltd, copied to lkml and various news sites."

Also stated in the article are the two areas which the problem has occured with. Namely, these are the PCI subsystem and the IO resource allocation, which has been mentioned in passing on the ARM Linux Mailing List.

If the accusations are true, and Castle have used the GPL Linux code, they will have two main options. The first is to release the source to the sections where the GPL code is used and the second is to rewrite the sections without using the GPL code. The first is by far the easiest, but making RISC OS (or parts of it) open source has been discussed before by RISC OS Ltd, who said that it would not be released as open source.

IconBar.com will keep you up to date on this as we get more information.
 

  Castle break GPL?
  This is a long thread. Click here to view the threaded list.
 
SimonC Message #91685, posted at 01:02, 8/2/2003
Unregistered user Withoyt any further information other than plain allegations, what I've read on Slashdot sounds like typical Slashdot ignorance and jumping on the "Linux & GPL issues are always the way we want them" bandwaggon.

I'm stretching myself a bit further here, not having bothered reading through the GPL, but surely the whole point of open source is that people can use it? Have Castle been both proven to use GPL code and not provide the source they used that derived from GPL code? Until either of those comes up it should be a case of "innocent until proven guitly", and applying that universally seems to be beyond most Slashdot readers.

  ^[ Log in to reply ]
 
Daniel Barron Message #91686, posted by danielbarron at 01:27, 8/2/2003, in reply to message #91685
Member
Posts: 19
Just read the GPL - its not that complex!

Also GPL != Open Source. Open source just means you provide the source which can be under any license. GPL is open source as is the BSD license and others. The whole point of the GPL is to keep the work and anything dervied from it in the public domain.

One of the best comments I saw asked why they did not use netBSD code instead as its license basically says "do what ever you like". NT and 2k have BSD code as does the RISC OS TCP/IP stack IIRC.

If the code is really embedded as opposed to just being a module then the entire program source is required to be released.

  ^[ Log in to reply ]
 
Drew Message #91687, posted at 01:42, 8/2/2003, in reply to message #91686
Unregistered user I've a couple of concerns here. Firstly, RISC OS <> Castle, which is a misconception bred upon the Slashdot site and indirectly repeated here and on other sites, and isn't helpful. Russell should know that and should certainly know better than to have phrased it in this way, particularly given the extremely hazardous payload of his e-mail.

Hence, as I see it, Castle has zero option about whether it can release 'RISC OS' source - that right lies with Pace, and to 'buy' sections to open source would be a costly business! It could, of course, release its new PCI manager and I/O subsystem code - but if you include GPL components with non-GPL components IIRC you risk your license too, though don't quote me on that. There is now also a question of whether this is Castle's branch of the RISC OS code which contains this code, or whether it is Pace's branch. The two are different ballgames.

I do find it amusing though that the same people that are advocating anti-RIAA laws so that they can't copy their CDs and DVDs are the first lining up to act on rumour!!! <mutters>Of course, that's purely for 'personal use' and it's totally different, mumble mumble</mutters>

  ^[ Log in to reply ]
 
Darren Winsper Message #91688, posted at 13:56, 8/2/2003, in reply to message #91687
Unregistered user First of all, it's a little more than a rumour, it's coming from a well respected Linux kernel developer. Second of all, no matter where it comes from, Castle Technology have no right whatsoever to incorporate GPL code into RISC OS without making RISC OS GPL. Thus, they must cease shipping any offended code immediately, or they could find themselves in court.

The real irony is, if the EUCD had been incorporated into UK law by the deadline, Castle could find themselves up on criminal charges.

  ^[ Log in to reply ]
 
Annraoi Message #91689, posted by ams at 17:46, 8/2/2003, in reply to message #91688
Member
Posts: 56
Oh I see, so because the rumour comes from "a well respected Linux kernel developer" this means that it must be true, right ?

Hey let's suspend any form of critical assesment of that it must be true..... ;) [Heavy Sarcasm has now ceased]

Regards

Annraoi

  ^[ Log in to reply ]
 
Chris Williams Message #91690, posted at 18:47, 8/2/2003, in reply to message #91689
Unregistered user There's evidence by Justin Fletcher, linked to from drobe.co.uk. He looked at the RISC OS 5 ROM image, saw some familiar function names embedded in it and investigated. Take a look.

Chris.

  ^[ Log in to reply ]
 
anon Message #91691, posted at 19:37, 8/2/2003, in reply to message #91690
Unregistered user So, can we have a GPL'd copy of RISC OS 5 for RedSquirrel then please?

So that's why there's been so many new versions since release....

  ^[ Log in to reply ]
 
Winston Message #91692, posted at 23:38, 8/2/2003, in reply to message #91691
Unregistered user The point about the GPL isn't that the source is Free, is that if you use that source in something you write, you must also make your code GPLed too - i.e. give back to the community who you based your code/product on.

To comply with the GPL, Castle would have to release the source for all the parts of RISC OS which are linked with the GPLed code, which would probably be a substantial part of Risc OS. If they don't do this, they have no right to use Linux kernel code in their product. Unlike the BSD licence, which allows you to take and not give back, the GPL requires you give back - in the form of source code. From the sounds of it, they aren't acting honourably.

There is also the LGPL which is typically used for libraries, so you can develop a closed-source product using an LGPL library without having to publish your source code. However, no Linux kernel code is licenced under the LGPL.

  ^[ Log in to reply ]
 
ecf Message #91693, posted at 07:07, 9/2/2003, in reply to message #91692
Unregistered user If Castle had provided a separate module in its RISCOS package using an exact copy of the source code without modification then Castle could be guilty of not attaching a copy of the GPL to it and not attributing its source. On the basis of what has been said about the purposes of the software I see no reason why GPL software could not be used quite properly in a RISCOS distribution.
If the software is completely separable e.g. as a binary module, and Castle had modified the GPL source then the source code of only that particular modified software module would need to be provided and not the whole of the RISCOS version/distribution.
Commercial entities are allowed to charge a fee for providing copies of GPL software. If it is proved that Castle charged more than just a handling/copying fee then it would be breaking the GPL terms.
Let us get this into proportion. If the GPL software can be used properly for RISCOS developments then it is to the benefit of the RISCOS community. Keeping RISCOS going commercially is a monumental task and it needs every assistance to survive.
  ^[ Log in to reply ]
 
Darren Winsper Message #91694, posted at 12:42, 9/2/2003, in reply to message #91693
Unregistered user ecf, you're wrong. Here's a simple acid test; take out all GPL code from RISC OS 5. Does RISC OS 5 still function? No, it doesn't. Thus, it would be RISC OS 5 itself that would have to be GPLed. What you are describing is exactly what the FSF didn't want to happen.

And why are you just thinking of the RISC OS community in all this? This violation doesn't help the Linux community, the people Castle (are alleged to have) stole from.

  ^[ Log in to reply ]
 
John Hoare Message #91695, posted by moss at 13:23, 9/2/2003, in reply to message #91694

Posts: 9348
Well, you would rewrite bits of RISC OS 5 so it *does* function without the allegedly stolen code (last time I checked, my Risc PC operated without it ;)). *Then* you would release a module that implemented all the PCI stuff, etc, and include the source.

Simple.

  ^[ Log in to reply ]
 
Moose Message #91696, posted at 14:27, 9/2/2003, in reply to message #91695
Unregistered user Your RiscPC doesn't have PCI though!
  ^[ Log in to reply ]
 
Darren Winsper Message #91697, posted at 14:29, 9/2/2003, in reply to message #91696
Unregistered user moss: But you can't do that. RISC OS 5 would not function in its entirety without the (alleged) GPL code. See section 3 of the GPL:
For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable.
Last time I checked, RISC OS 5 would be considered an executable work, seeing as it's executed upon boot-up.
  ^[ Log in to reply ]
 
John Hoare Message #91698, posted by moss at 14:41, 9/2/2003, in reply to message #91697

Posts: 9348
Hmn. But if previous RISC OS versions do work without GPL code, then it can hardly be called an integral part of the system - so shoving it in a module is fine; but, then, I suppose it depends what you mean by integral. If the graphics card isn't working, so you can't make any use of the system, perhaps that *does* count as integral.

/me confused :-/

  ^[ Log in to reply ]
 
John Hoare Message #91699, posted by moss at 14:42, 9/2/2003, in reply to message #91698

Posts: 9348
I know my RPC doesn't have PCI, Moose - but it's all a case of whether PCI stuff counts as integral, doesn't it. I had decided it wasn't; perhaps now I'm not quite so sure.
  ^[ Log in to reply ]
 
Darren Winsper Message #91700, posted at 14:48, 9/2/2003, in reply to message #91699
Unregistered user Look, just read the GPL and the GPL FAQ. Pay careful attention to this section:
http://www.gnu.org/licenses/gpl-faq.html#TOCMoneyGuzzlerInc
  ^[ Log in to reply ]
 
Moose Message #91701, posted at 16:11, 9/2/2003, in reply to message #91700
Unregistered user It's quite simple - without PCI, an Iyonix won't do /anything/. It's 100% integral.
  ^[ Log in to reply ]
 
SimonC Message #91702, posted at 19:32, 9/2/2003, in reply to message #91701
Unregistered user I don't buy that argument, and I think if the GPL insists on it then it might not stand up in court. Imagine another machine that uses RO4, but requires an extra module, which may have been put into the ROM, or may have been softloaded on boot up, which uses GPL code. I wouldn't call it integral, even if it's necessary, since it is a clearly distinguishable, separable piece of code.

The same file clause probably wouldn't hold either. In the case of a ROM image there are several different bits of software that are definitely not integral (such as Draw and Paint). You could argue the same with archive files.

  ^[ Log in to reply ]
 
Moose Message #91703, posted at 19:49, 9/2/2003, in reply to message #91702
Unregistered user Since the hard drive is a PCI device, where will we load this mythical from?
  ^[ Log in to reply ]
 
Gerph Message #91704, posted at 20:28, 9/2/2003, in reply to message #91703
Unregistered user SimonC: We're not talking about a module. This is in the kernel. That is, the during the machine initialisation the PCI system is started. The graphics and sound system run off it. It's pretty integral.

However, CTL say they'll be putting out a statement which will clear things up on Monday and then I'm sure we can all shut up because they'll be cleared completely of any wrong-doing.

  ^[ Log in to reply ]
 
Moose Message #91705, posted at 20:53, 9/2/2003, in reply to message #91704
Unregistered user Unless the statement is "Yeh, we ripped the Linux kernel. Come sue us if you dare. Ph33r 0ur O.S."
  ^[ Log in to reply ]
 
Darren Winsper Message #91706, posted at 21:32, 9/2/2003, in reply to message #91705
Unregistered user SimonC: I don't know how to address this other than saying "yes it ******* does". Seriously, read the GPL and the GPL FAQ. If you put GPL code in your product, you must distribute the entirety of your product under the terms of the GPL.

As for whether it'll stand up in court, why wouldn't it? The GPL grants more rights than standard copyright law. If they didn't agree to the GPL, then they could ship their product at all.

  ^[ Log in to reply ]
 
Darren Winsper Message #91707, posted at 21:48, 9/2/2003, in reply to message #91706
Unregistered user Oops, make that "couldn't ship their product at all"
  ^[ Log in to reply ]
 
Jeffrey Lee Message #91708, posted by Phlamethrower at 22:46, 9/2/2003, in reply to message #91707
PhlamethrowerHot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot Hot stuff

Posts: 15096
And there was me thinking that RISC OS 5 & the Iyonix where the 'right' direction to go in :o
  ^[ Log in to reply ]
 
scarlet_pimpernel Message #91709, posted at 00:49, 10/2/2003, in reply to message #91708
Unregistered user I can't believe such a fuss is being made based on the allegations of two sad nerds. Just who would be SO sad as to go round looking at ROM images and the like?

I find it too coincidental that one of the accusers happens to work for RISCOS Ltd who have been effectively killed by RISC OS 5. It won't be long before Select subs dry up completely as people invest in the new technology, as opposed to throwing money at the RiscPC which has simply had it's day. Who would be daft enogh to keep subscribing to updates to a 26bit soft-loaded OS whilst a 32bit ROM-loaded version is being developed and run on better, faster hardware?

And as Castle and Pace are well aware of the number of pathetic nerds that RISC OS and Linux unfortunately attract, they would hardly likely do something like this in the full knowledge that some saddo like Justin Fletcher would look for it straight away.

I thought for a while that people would fail to find something major to whinge about over the Iyonix and 32bit. I mean, now it's here, the whingers have had their End Of RISC OS Due To No 32bit mantra crushed. If there's nothing wrong... make something up!!

It's no wonder they wound Acorn up. Excellent machines, fantastic engineers and a customer base consisting of nerds and whingers (half of which don't buy anyway).

Castle have saved RISC OS. The least you whingers could do is allow them the legal and moral right of 'innocent until proven guilty'.

  ^[ Log in to reply ]
 
Chris Williams Message #91710, posted at 01:57, 10/2/2003, in reply to message #91709
Unregistered user Aw, poor baby. Thanks for the rant dude. It's these "nerds" that develop the software and systems you're using right now. Justin working on RISC OS 4 (and thus the bits of the OS that made their way into OS 5) and rmk (if the allegations are proven true, of course) working on the PCI code that helps drive the Iyonix. So pardon them for protecting their work or just being naturally inquisitive.

Castle _are_ innoccent until proven guilty. That's why we're waiting patiently for their side of the story this week. So sit down.

Chris. Jesus, zealots.

  ^[ Log in to reply ]
 
Horse Message #91711, posted at 10:03, 10/2/2003, in reply to message #91710
Unregistered user I can imagine that this has already been said, but what is Castle's business plan exactly?

1. Violate the GPL.
2. ???
3. Profit!

:-)

  ^[ Log in to reply ]
 
Andrew Message #91712, posted by andrew at 11:12, 10/2/2003, in reply to message #91711
HandbagHandbag Boi
Posts: 3439
I think you're right Scarlet - it sounds like a malicious smear on Castle.
  ^[ Log in to reply ]
 
Horse Message #91713, posted at 11:18, 10/2/2003, in reply to message #91712
Unregistered user Oh "boo hoo", accompanied by the sound of stamped foot! World domination can't even be achieved by violating the GPL. It's a sad day for the Acorn legacy, but have you not realised why?
  ^[ Log in to reply ]
 
ecf Message #91714, posted at 11:28, 10/2/2003, in reply to message #91713
Unregistered user "Oscar Wilde's dictum that there is only one thing worse than being talked about and that is not being talked about finds its modern-day rendering in the oft-repeated claim that all publicity is good publicity." I quote from the Independent yesterday, but sadly the article was not talking about RISCOS, but it could well have been. How many Slashdot readers had ever heard about RISCOS before this incident? Maybe a few will be interested enough to find out what RISCOS is all about.

However, I believe it is unwise for individuals to make public statements that could in any way be construed as damaging to the reputation of a person, organisation or company unless a) they are absolutely sure of their ground and can prove the truth of it and b) have deep pockets to defend themselves. The following website gives a quick explanation:

http://www.rjw.co.uk/legal/libelxpress/defamation.aspx

[ Incidentally "RISCOS" finds significantly more content on Google than does "RISC OS", so don't buck the trend.]

  ^[ Log in to reply ]
 
Pages (2): 1 > >|

The Icon Bar: News and features: Castle break GPL?