RiscOS/BBC - TCP/IP
Jess Hampshire (158) 864 posts |
Would it allow GPL modules to live within resourcefs? |
Steve Pampling (1551) 7931 posts |
GPL items in a filing system along with non-GPL certainly fits. Whether FSF people can get their head round ResourceFS without their brains swelling and exploding is another thing. As an aside, if someone produces an application (closed) and only puts lic/copyright info in limit source files then later declares that everything with a declared lic. is now GPL – how much is GPL? |
Dave Higton (1515) 3404 posts |
If I’ve understood you correctly, that looks like dual licensing. It isn’t possible to apply or change a licence retrospectively. |
Steve Pampling (1551) 7931 posts |
That was my view.
That’s my belief too. |
Tony Noble (1579) 62 posts |
Out of curiosity, is there anything preventing distribution of the 26 bit executable + binary patch, until the permission issue is resolved? |
Steve Fryatt (216) 2045 posts |
whatever version of SQL they were using seemed absolutely unable to reliably cope with beeping in stuff on-site Really? I’m not sure how the specific SQL implementation actually affects what Rick was claiming: it’s the end application that’s broken, not the underlying database engine. Unless you’re trying to suggest that we can blame BBC BASIC for each and every duff application that’s written using it? |
Steve Fryatt (216) 2045 posts |
It depends on what Justin originally said (and I don’t have that info to hand). EasySockets appears not to be on an Open licence, so if Justin said it couldn’t be used on 32-bit systems, that’s it. |
Steve Pampling (1551) 7931 posts |
Giving the other side of the coin – an application I have in use at work: |
Frank de Bruijn (160) 224 posts |
From the !Help file in the EasySockets 1.18 archive: Use of this module Disclaimer |
Steve Fryatt (216) 2045 posts |
If I’ve understood you correctly, that looks like dual licensing. Which is legitimate, no? It isn’t possible to apply or change a licence retrospectively. Is that actually correct? The copyright holder can do what they like, surely? There’s certainly nothing to stop me taking an application that’s entirely my own work and already released under a restrictive licence (perhaps “freeware”), and re-releasing it (changed or otherwise) under an Open licence. I own its copyright, so I can do what I like with it. If not, forget all of the work that I’ve done migrating my code over to the EUPL during the past few years. I’m not sure that there’s anything to say that I can’t do the reverse, either. If I take a piece of my own work, that’s licensed under an Open licence, I can always release a different version under a more restrictive licence. I can’t remove the licence that’s already out there, but while CashBook 1 is EUPL, I could quite happily make CashBook 2 anything I choose. The difference here is that I couldn’t stop you, or anyone else, taking on the distribution of CashBook 1 as the public versions of that will always carry the license that I put them out with. I could also release CashBook 1 under a different licence, but if it’s more restrictive than the EUPL there’s no point if there are copies of the EUPL version out in the wild. The problem comes if CashBook 1 starts to include someone else’s code: an update or bug fix that they submitted to me while it was licensed as EUPL, perhaps. I don’t have the copyright to that code (because its author does), so I then can’t change the licence of the application as a whole without their agreement – unless I remove their code completely. Even if done legitimately, I could see that idea getting messy. The tricky bit is what happens if I decide to stop distributing my software. The GPL would require that I made the sources available for a period of time after this; some other licences don’t care. If someone else has the source, they can continue to distribute it, and updated versions, under the licence that I chose (but then can’t change the licence, as they don’t own the copyright to the bulk of the code). If no-one downloaded the source, there might be problems distributing the binary under the terms of some of the Open licenses (as you often can’t send someone the binary if you can’t deliver on the source). Which is why, if you’re using Open Source software, it’s always handy to have a copy of the source – even if you never plan to do anything with it yourself. That said, I should probably clarify that I’m not contemplating changing the licence used by any of my software (unless it’s still “freeware”, in which case it might change to EUPL at some point in the future). [Etited to fix some nasty syntax] |
Steve Fryatt (216) 2045 posts |
MySQL and MS SQL aren’t really compatible (in the same way that ARM BASIC and BB4W aren’t). |
Steve Fryatt (216) 2045 posts |
Oooh. Now there’s a thing… While “freeware” and “shareware” are pretty meaningless terms, I was under the impression that “Public Domain” isn’t. If that’s out there in an archive with the source as released by Justin, is he even in a position to say that he doesn’t want versions made available for the Iyonix?
Ah. A contradictory licence. Oh well. |
David Feugey (2125) 2687 posts |
It is possible, but hard (see Linux change from GPLv2 to GPLv3).
No, I suggest the opposite. Implementation was probably the problem.
Absolutely. GPL is not a jail :) |
Steve Fryatt (216) 2045 posts |
Implementation of the end application, not of MySQL. |
Steve Pampling (1551) 7931 posts |
No, that’s a change of stance for all new code and an existing situation for the existing code. Any attempt to add retrospective restrictions is the usual FSF bull. If it’s out and open under condition 1 then that version is out and open under condition 1. The next, amended very minorly, version can be restricted more (or less). 1 Most files actually contain text to say that the stuff at the top must remain as is. |
Rick Murray (539) 13401 posts |
No. The only way such a licence can be retrospectively changed is if a judge rules the application of the original licence was invalid. More later. Hard to write stuff at break on a phone while stuffing noodles in the cake hole. ;-) |
David Feugey (2125) 2687 posts |
Exactly my point (and words).
Of course, prior GPLv2 version will still exist, but you can change the license of your code and say ‘it was GPLv2, it’s now also GPLv3’. So new versions will be completely GPLv3, not only the new code (even if old code is available in two versions: GPLv2 & GPLv3). The problem is that normally you simply change the licence. With GPL you must also agree to relicense existing/previous code. That was the problem with Linux. |
David Feugey (2125) 2687 posts |
Example: TrueCrypt. |
Steve Pampling (1551) 7931 posts |
For all future code written after the declaration.
FSF bull. History is history, attempting to re-write it is hubris. Which I suppose sums them up. |
Steve Fryatt (216) 2045 posts |
Implementation of the end application, not of MySQL. Maybe there’s a translation problem here? Your words were:
which says exactly the opposite of what you’re now backpeddling towards (if MySQL is OK, why describe it as a “mistake” or suggest changing to use PostgreSQL)? |
Rick Murray (539) 13401 posts |
Before we start – this post is like one of those Japanese movies where time is anything but linear. Just so you’ve been forewarned… Steve: It isn’t possible to apply or change a licence retrospectively. Depends upon the licence. If I use one of your EUPL programs and you later decide to close-source it, I can continue using the EUPL version regardless of whether or not you want me to. Because the EUPL (that you placed the code under) says I can. David:
Yes. ONLY the new code; or old code downloaded from that date. Older code that you already have will be either GPLv2 or GPLv3 (your choice – if the copyright holder gives that option). Why? This at the top of each piece of source:
Note the final seven words. If I have a copy of the program which says exactly that, and it is later re-released as GPLv3, there is nothing FSF or the GPL advocates could do to change my copy. I can either download the updated source (accepting GPLv3), decide to accept GPLv3 with the current source, or continue with GPLv2 (including releasing my own sources as GPLv2). Why? Because the licence grants me that right. David again: It isn’t possible to apply or change a licence retrospectively. Well therein lies a story. The FSF traditionally asks for copyright to be assigned to them to provide them with a means of enforcing the GPL. Since this is a headache for individual authors, a fair number of people have assigned their copyright in various GNU tools to FSF. Thus, Linux may itself be in contravention of the GPL – there are no exception clauses for using software with differing versions of GPL – each licence goes all Highlander and shouts “THERE CAN BE ONLY ONE!”. More David:
? Jess asks:
and David said:
I beg to differ. Let’s examine Jess’s request. One would nominally say that it is acceptable to distribute GPL and non-GPL programs on a storage medium together (the primary problem seems to be what links with what and how). The thing is, this isn’t an ISO image. This would be a ROM image. A prebuilt binary blob that encompasses “all” of RISC OS and may have a GPL component within. The GPLv3 says this:
and (my emphasis):
and:
While section 5 refers to distributing modified source versions, it very much counts in the case of a binary only release as section 6 says that the object code must also be accompanied by its source code (in various manners). So – tell me – where exactly does all of this linkage rubbish derive? If a GPL program is part of ResourceFS which is part of RISC OS does it count as an aggregate or would it imply that everything should become GPL? The program (however it is written) would have to use no part of RISC OS (modules could count as “shared libraries” for section 1; though there is a risk of all sorts of potential confusion as RISC OS architecture is very very different to Linux).
David:
Unfortunately the FSF have been rather clever here. While we (Europeans) can pretty much ignore the entire GPL FAQ and all the illogical FUD within it (if it isn’t specifically stated in the licence text, you can’t be held to it); under American law a judge can choose to seek guidance for the intent of the licence, which means that the FAQ may (at his/her discretion) carry legal weight. It’s a scary thought… So, David. You think the GPL is a clear licence? Please authoritatively answer Jess’ question. Can a GPL program be included within ResourceFS (and distributed with RISC OS) without the GPL affecting any other part of RISC OS? For my part – I would say the only safe option is to include such a thing separately (otherwise known as “not including it”!). Why? Because the limit of the “contamination” of GPL is actually extremely unclear. In fact, you either have to know this through osmosis (because it is just one of those things about the GPL that everybody is aware of in some degree) or you have to read all sorts of terms and conditions that these days seem to be more preoccupied with patents and anti-circumvention restriction. Now read this: http://opensource.org/licenses/CDDL-1.0 You will notice that it retains CDDL on CDDL files, but interacts and co-exists with other licences. By contrast the GPL is extremely unfriendly in that it only interacts with itself (the “open at any cost” approach). Furthermore, as there is no attempt by CDDL to pollute any other code by mere proximity, there is no need for random vaguely-worded statements regarding what happens with other parts of the project. Indeed, section 3.6 explicitly states:
The openness of CDDL code remains open. The rest? Not CDDL’s problem. Do you see the difference? This isn’t about whether or not the GPL viral behaviour is good or bad (I pick this option!), it is about the fact that the GPL is an extremely popular and widely used licence that contains several extremely important caveats that are ill-defined in an appallingly written licence. It is as if the authors could envisage nothing beyond the cosy “everything is free!!!!” world of GNU/Linux and don’t care about crusading for everything to be GPL as they themselves are the driving force behind GPL. The GPLification of the world benefits them and their cause. [To give you an idea – the FSF notes that it is possible to subvert the EUPL (which does not list GPLv3 as a compatible licence) to become “compatible” with GPLv3 via a roundabout route of using another licence – WTF? http://www.gnu.org/philosophy/license-list.html#EUPL ] The GPL is just not clear. Period. My dislike of GPL derives exactly from this point alone. I don’t care much for the viral approach but I can choose whether or not to use GPL code. Or to put it in much simpler terms – I’m a programmer, dammit, not a lawyer. I want a licence that has clear obvious terms that everybody can understand and decide whether or not it suits them. My preferred licence is the EUPL, I have no issue with CDDL either. I don’t want smoke and mirrors and big egos. That’s what the GPL offers. And, ironically for a movement all about freedom of source code, the GPL offers a lock-in to a GPL world where said freedom is quite narrowly defined. Oh, and that whole SQL thing? I would imagine some sort of sub-contractor did it rather than in-house. Okay. I’ll get off the soapbox now. Time for somebody else to take a turn… ;-) |
David Feugey (2125) 2687 posts |
No, also for the existing code, with agreement of all contributors. But of course, existing code will also be available under the old license TOO. Dual licensing is the only solution. If existing GPL code is not relicensed, future code will not be able to use another license, since GPL is viral. Dual licensing is possible at every moment. Good example is Qt. Was 100% QPL, then 100% GPL, then 100% LGPL. In Qt 4.5, there is no 95% GPL and 5% LGPL (IE 100% GPL, since viral). It’s all relicensed as LGPL.
That was sarcastic. Look at ‘and a good database manager’ (the guy who makes the implementation). Very often, when a company uses SQL Server and chooses to go to open source, bad project managers choose MySQL because it’s popular… It will solve all problems anyway. Bad choice. 1/ you need good a project manager. You also tell: you need a good project manager (who will choose the right product). nota : “and sold them a Microsoft-based solution” Long version: since the Microsoft solution worked, I understand that 1/ the project manager was good (‘and a good database manager’) 2/ he used an ORDBMS (SQL Server). Perhaps because it was needed? So my remark on MySQL (RDBMS) VS PostgreSQL (ORDBMS). |
Steve Pampling (1551) 7931 posts |
For all future code written after the declaration. I think you’re changing position. You’re now adding that all contributors agree as opposed to just unilateral change. It’s wrong anyway. The old code has a licence in the source that can’t be re-written for that copy. You could try and people like me would respond with a Foxtrot Oscar. Actually, mostly I’d do a good imitation of a deaf Scarlet Pimpernell. (You wouldn’t find me and if you did I’m not listening.)
What makes you assume that? MySQL, PostgreSQL, MS SQL all exist on MS platform. |
David Feugey (2125) 2687 posts |
No. Absolutely not. See Qt or TrueCrypt examples. “It’s very simple: If you own the copyright, you can relicense it as you want. Period. http://ubuntuforums.org/showthread.php?t=1322006 “Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice.” (here system = distribution method described by GPL license) GPL code will stay your code. Of course, contributions are property of contributors. |
Steve Pampling (1551) 7931 posts |
Oh, absolutely yes. You cannot re-write history. It was what it was. It is written. |