LWN: Comments on "The Party of Gno (Linux Magazine)"

LGPL-like license

farnz — Wed, 30 Jun 2010 20:50:50 +0000

It's also possible for the closed source application to do a partial link; take the bunch of .o files that make up the closed source application, link them together with appropriate options (see GNU ld's --relocatable option, if that's your linker of choice), and strip as much of the symbol table as possible. The resulting .o file gives away very little about the original closed-source code (on a par with a closed source app using the LGPL libraries as shared libraries), but is still good enough for linking against many other .o files (those from a LGPL library, for example) to get a final binary.

LGPL-like license

foom — Wed, 30 Jun 2010 19:25:14 +0000

> the only way to do this with the LGPL is to make a shared library (so that the end user can, supposedly, replace it with a new version)

That's not true at all. I don't believe shared libraries existed (or at least, they weren't yet common) when the LGPL was introduced.

LGPL: 1991.
ELF: 1992?
ELF on linux: 1994.

The accepted way to follow the LGPL without using shared libraries is for the closed-source program to distribute the .o files you can pass to a linker, along with the .a library, to create a final executable. I understand that you don't want them to have to do this, but that's different from saying it's impossible or not allowable.

As the LGPL says:
> Accompany the work with the complete corresponding machine-readable source code for the Library including whatever changes were used in the work (which must be distributed under Sections 1 and 2 above); and, if the work is an executable linked with the Library, with the complete machine-readable "work that uses the Library", as object code and/or source code, so that the user can modify the Library and then relink to produce a modified executable containing the modified Library. (It is understood that the user who changes the contents of definitions files in the Library will not necessarily be able to recompile the application to use the modified definitions.)

LGPL-like license

spitzak — Wed, 30 Jun 2010 00:08:21 +0000

I think I might have been misleading as to what I wanted to do. I am not shipping .a files or anything else. I want to ship the *source* for my software, but allow somebody to use my source in a closed-source program, provided they do not modify the source files themselves, but only add other files to them. Most likely they will compile my stuff to a .a file but they will have no trouble making a new .a file if they need to, and what they distribute will be an executable, not a library.

What I am looking for is a short name and legal description, blessed by the FSF, for what I have been calling "GPL with linking exception".

LGPL-like license

nix — Tue, 29 Jun 2010 21:02:14 +0000

No. glibc makes no guarantee of binary compatibility except on the level of ELF executable objects (that means executables and shared libraries, which are very nearly identical beasts in many ways with different things conventionally done to them). Specifically, there is no guarantee at all for .o files, and that means for .a files too. The only .o files you normally see on a Linux system are a few startup/shutdown files in /usr/lib which come with glibc (and so the lack-of-guarantee is irrelevant) and GCC (and these contain no glibc references).

That is, you cannot compile against one version of glibc and link against another, ever. That's not what the interface guarantee allows. What you can do is compile and link against one version and then *dynamically* link (i.e. *run*) against a later version, safely, even if functions change in ways that might disturb your program or structures like 'struct stat' change size.

There are two reasons for this.

Firstly, .a files cannot make reference to specific versions of versioned symbols: that work is done by the linker. As most of the glibc interface-compatibility maintenance scheme depends on versioned symbols, this means that scheme cannot operate for .a files at all. Also, there are a good few functions in glibc where the old version depends on the header files the code was compiled with doing one thing, and the new version depends on the header files doing some other thing. Compile against one version of glibc and link against another, and pain results.

Secondly, glibc depends (because it must) on determining the version of (potentially versioned) structures bound into a program by means of linking a .a file -- libc_nonshared.a -- into that program and having your program pass a function in that file a version number tag defined in the headers at compile time. Compile against one version and link against another, and the tag may be wrong (hell, the structure may have different sizes in different parts of your program: obvious disaster). (See http://sourceware.org/glibc/wiki/Development/Versioning_A... for much more information.) (This one will not have bitten you so far, because none of these versioned structures have ever changed definition yet. But they might, and if they do, little things like stat() will rely on them.)

So you can ship .a files if you like, as long as they never use any functions in glibc that take any structures as parameters whose definitions might ever possibly change, and never call any functions from glibc which might ever be versioned (that would be all of them). I hope your useful functions never use glibc. ;)

Note that you don't have to finalize the API. You just have to *make a shared library*. You can give it a useless soname -- increment it whenever you release your program, soname it after the date, anything -- and change the API whenever you like, and laugh at anyone else who DT_NEEDs it, but you cannot ship .a files and expect them to work, at all, across glibc upgrades. The X people used to do as you do -- experimental extensions got .a files only, so they didn't have to give the shared library an soname -- and glibc upgrades broke them several times as I recall. Eventually X.org shifted to using shared libraries for everything as part of the modular rework, and just using an soname of {blah}.0 until the interface stabilized. It was a good decision.

LGPL-like license

spitzak — Tue, 29 Jun 2010 18:05:11 +0000

Distributing static libraries is bad for a vast number of reasons beside the lack of easy LGPL support; notably, you lose all backward-compatibility guarantees for glibc, so you need to ship a new version for every glibc minor version.

Huh? This would also apply to programs. You are saying that libraries must be recompiled for every minor change to glibc, but for some reason programs don't?

I think possibly the confusion is that you think I am talking *about* glibc. I am not. I am talking about MINOR libraries of useful funcitions, used by approximately ONE program on a system. A shared library is a waste of time, and is counter productive to me. And I find it hard to believe you think I should be required to finalize the API before I let anybody test my software!

LGPL-like license

nix — Tue, 29 Jun 2010 04:43:11 +0000

Not only does this make it a pain to distribute the application, a much more odious problem is now I am forced to be binary-compatible

As you pointed out in the next line, no you aren't. There's nothing stopping you setting the soname to .0, or to the program version, and just dumping binary compatibility (although if a library interface is changing *that* fast once it's in active use and it's not a compatibility library like gnulib, I'd have to wonder 'why'?)

This is not exactly rocket science. Hell, it's how half the shared libraries on your average Linux system are 'versioned' (the ones that are associated with only one package).

LGPL-like license

spitzak — Tue, 29 Jun 2010 01:18:43 +0000

I have found the LGPL to be a pain for developing open-source libraries.

On the assumption that I actually want to encourage the use of my library by allowing closed-source programs to use it, the only way to do this with the LGPL is to make a shared library (so that the end user can, supposedly, replace it with a new version). Not only does this make it a pain to distribute the application, a much more odious problem is now I am forced to be binary-compatible, which I certainly do not want to do for an under-development library (in reality I don't bother with binary compatability but this means that nobody can actually replace it in a program anyway, thus defeating the entire purpose of this LGPL requirement).

The FSF does support licenses like I want, they typically call it "GPL with a linking exception". What I would like is a SHORT name for this, and official legalese code. The lack of this is probably the main reason for OSS license proliferation because everybody writes their own.

Unlike the CDDL and some others, I want to allow my code to be used by anybody including GPL projects. And I like things like the reverse-engineering clause, etc, from the GPL, which are often deleted as well. I just want end users to be free to write any program using my code and distribute it any way they want, while being unable to make a closed-source modification to my code.

LGPL-like license

foom — Sat, 26 Jun 2010 04:09:26 +0000

Huh, I always thought the LGPL was to protect the users, by allowing them to fix bugs in the software.

It thus requires that you allow the replacement of the LGPL-licensed work in your application with a newly compiled version. Makes sense to me...

Now, a name for the license you want which abandons the users would be nice, but I would've omitted the "what people think the LGPL is" bit of your comment. :)

LGPL-like license

spitzak — Fri, 25 Jun 2010 23:38:21 +0000

Can you tell me if the FSF recommends or has defined a license that is basically "what people think the LGPL is". I would dearly love to be able to stick a TLA on my software and have everybody know what it is, rather than the current kludge of saying "GPL + linking exception".

Just to be clear, what I am looking for is that somebody can make a program using the licensed software in any way, as long as they don't modify the software itself, and do anything with that result, such as sell it without source. However if they *modify* the software, they must include the source code of their modified version (they can then make a closed-source program using the modified version). There may be a few other rules to avoid tricks: mostly that the modifications must be stand-alone and result in a piece of software of equal reusability as the original. In addition you must be able to combine the software with GPL/LGPL licensed software, in which case the result is GPL/LGPL, this requirement makes the CDDL unuseable.

The Party of Gno (Linux Magazine)

pboddie — Thu, 24 Jun 2010 13:38:55 +0000

We have spent a century in the west making people who care about liberty of any sort into wierdos. You aren't supposed to want to be Free, you are supposed to want to be taken care of. And Apple, Facebook, Google, Microsoft, etc. are all too willing to 'take care of you.' as long as you don't inquire too closely at the true pricetag. And nobody is in the habit of looking gift horses in the mouth these days.

So true. It is, of course, shocking that people will criticise others for campaigning for various freedoms, effectively campaigning themselves to uphold the "freedom" of the corporate world (and others) to treat people in any way they see fit as long as the goodies keep coming. Indeed, people who are willing to surrender their statutory rights often appear to want to pressure everyone else to surrender theirs in order to keep everyone "in line" and not risk incurring the wrath of those supposedly generous and benevolent corporate masters.

Thus, when discussing things like product lifetimes and manufacturer obligations towards the customer, some people will actively criticise others for favouring things like a reasonable warranty period because "it might mean we don't get the new stuff as quickly" and "manufacturers might not want to sell stuff here any more". Suddenly, the threat of being cut off by their corporate masters drives them into action like nothing else.

But even if some people cannot be persuaded to reshape their relationships with the companies they do business with, the situation is not as bad as it could be. In many countries there are fairly strong "consumer rights" movements, and even if people have been conditioned into not thinking about freedoms, they have been taught about demanding a fair deal when putting down their money. All the FSF and others need to do is to bridge the gap between the work going on in such areas and the arguably less "mercantile" nature of their existing campaigns.

The Party of Gno (Linux Magazine)

jmorris42 — Thu, 24 Jun 2010 04:50:26 +0000

> Intellectual honestly forces me to consider the possibility that
> he's warning us against serious dangers.

Oh I guess about 90% of the people reading here agree the danger is real. The argument is about a) what can be done about it and b) tactics.

Personally I think the problem is bigger. Good luck convincing people to care about Freedom when it comes to their data on Facebook or their web browser's source code when they only pay lip service to freedom in general. We have spent a century in the west making people who care about liberty of any sort into wierdos. You aren't supposed to want to be Free, you are supposed to want to be taken care of. And Apple, Facebook, Google, Microsoft, etc. are all too willing to 'take care of you.' as long as you don't inquire too closely at the true pricetag. And nobody is in the habit of looking gift horses in the mouth these days.

And for all the fervor for Free Software here in nerdville, most aren't willing to see beyond this one case to examine the bigger picture. Any society that wants to be taken care of by someone else from cradle to grave doesn't really give a darn about Freedom. Raise yer hand if you oppose the rest of the nanny state too? You gave up the rest of your liberties for supposed security, why are you expecting to be able to pick and choose this one to keep?

Think of it this way. To borrow the example from the article of PETA, imagine how much harder their job would be if nobody even realized animals could suffer or had even bothered to think much about animals at all. And worse treated fellow humans cruelly and violently on a regular basis. (Imagine the world as it existed a thousand or so years ago.) They would have to start with the basic concept that cruelty and needless suffering in general was a bad thing before they could have a hope of bringing up the issue of cruelty to animals.

The Party of Gno (Linux Magazine)

kirschner — Wed, 23 Jun 2010 14:32:21 +0000

(I also posted this comment on my blog.)

[...] I fully agree with Joe's point that it is important to do positive campaigning. This is especially true in Europe. That is why we -- FSF's European sister organisation -- always tried to formulate our message in a positive way. Only with positive campaigning FSFE is able to be recognised by European politicians and in return get their recognition, e.g. with this year's Theodor Heuss Medal. But also the FSF already had positive campaigns. Here some recent examples from the FSFs:

Document Freedom Day: The Document Freedom Day is a day for Document Freedom and for Open Standards. FSFE did most of the work for the central organisation of the DFD in the past years to promote Open Standards. The FSF also has a campaign page for Open Standards.
rOgg On: For this years DFD FSFE promoted Ogg Vorbis. The German and the Austrian team encouraged two radio stations which already use Ogg vorbis, by giving them a prize and a tart with the "rOgg On" slogon. Deutschlandradio stated that they Feel more honoured than for the Grimme Prize". The picutres clearly show how positive that campaign was. During that campaign we translated FSF's PlayOgg website into German, which also is a positive campaign.
I love Free Software For this years Valentine's Day we started the "I love Free Software" campaign. People can show their love to Free Software.
PDFreaders.org FSFE's Fellows started a campaign for Free Software PDFreaders and also explaining Open Standards in this context.

There are more examples of positive activies on FSFE's activity page, and there are also more examples on FSF's campaign page. Although there are a lot of "Gyes" campaigns out there, the problem is to get media coverage with positiv campagins. Beside that sometimes you have to critise bad things. For example Digitial Restriction Management (DRM) is bad for software users. The FSFs work that users have to control over their computing. So we have to defend them and say "No" to DRM, even if we do not have a solution how to make money with music and videos in the future. It is important for the FSFs to keep a firm stand on users software freedom.

--
Matthias Kirschner

History editing

pboddie — Mon, 21 Jun 2010 11:05:06 +0000

They're not doing the right things. DBD is a complete and utter waste of time. Win7 Sins is a complete waste of time and FSF contributor dollars.

How are these campaigns a "waste of time"? They may not target the man in the street, but if they target a demographic that can then influence the man in the street, then they're worthwhile. I notice that at least one commenter in this discussion has referred to Windows 7 Sins as "FUD", but I now take that to mean "I don't like those people, and I can't think of a response" rather than anything which can actually detract from the message of the campaign.

Do you regard any kind of "consumer education" campaign as a "waste of time"? Why not just give the consumers something which is just as shiny but less "evil" so that they don't really have to think about any issues? This works up to the point where Apple, say, make a public relations play about how they've made a more "open" product, and then you're back to square one.

OSI, last I checked, wasn't seeking funds to squander to run pointless campaigns.

Last I checked, OSI had no direction whatsoever, and probably no inclination to execute any campaigns that might be deemed "ideological", in the fear that it may offend all those people who have worked so hard over the years to not have a position on anything other than maybe how great the resulting code is: a position which is actually easily undermined and largely tangential to the matter of openness and its principal, blatantly obvious benefits.

Indeed, due to the nature of many of the other organisations around "open source" - that they eschew what they regard as "ideology" in favour of not rocking anyone's boat - there are actually few of them who are inclined to pursue the causes that the FSF pursues. If one dislikes the nature of the FSF's campaigns, one would be better advised to improve the FSF's campaigning instead of lining up some "pragmatic" superheroes, only to see them issue a collective "meh" on whichever issue one would have them confront.

History editing

coriordan — Mon, 21 Jun 2010 01:05:36 +0000

RMS isn't paid by FSF, so how he spends his time is his business. He's been watching this issue develop for 27 years and he's come to the conclusion that terminology is critical.

Was Windows7Sins a flop? It wasn't my favourite campaign, but FSF were right to do *something* for the release of Windows 7, and maybe another demographic liked it.

Did it consume donations? I see a website, a mail-out of 500 letters, a morning spent throwing Windows boxes into a bin in a nearby park, and a call for donations if people want more such letters sent. Not exactly extravagant. How do you know the bulk of the work wasn't done by volunteers? Or that someone didn't suggest it and foot the bill?

If you think scratching off actions like Windows7Sins could free up resources to make a substantial start on web services, I think that running an NGO would deeply disappoint you :-)

(and as I pointed out above, FSF is working on web services.)

The Party of Gno (Linux Magazine)

yuhong — Sun, 20 Jun 2010 20:38:10 +0000

Yea, I was going to say something similar about artificial scarcity too.

History editing

dlang — Sun, 20 Jun 2010 19:32:00 +0000

you forgot to add GPLv2 to the open source group

Gno Thanks

coriordan — Sun, 20 Jun 2010 02:43:08 +0000

> I'm asking [FSF] to look forward and think about how to create free services for cloud computing the same way that it created a free replacement for unix. Even if FSF doesn't do all the work or even act as a focal point, it would be nice if the FSF had a vision for free software that extends to SaaS and other "cloud" computing.

FSF published the AGPL in 2002, the Franklin Street Statement in 2008, essays on javascript and SaaS recently:

Please give me fewer articles, and more diamonds

coriordan — Sun, 20 Jun 2010 02:05:05 +0000

I think we all (here) agree that it would be great if the success of GNU could be replicated (ASAP) for web services. But GNU wasn't an overnight success, and it still has to fight constantly to even get recognised. Here's a 1986 speech about why we need software freedom:

http://www.gnu.org/philosophy/stallman-kth.html

This would fall under your "proselytising", which you seem to view negatively. In 1986, there was no GCC, GDB, glibc, etc. etc. Richard worked on it, and published essays, and built support, and he coded along with those supporters, and 27 years later here we are.

He's doing the same thing again (although this time seeking programmers to help from the outset), and we're in the early years, and there's no big success yet. You can either help out, or you can be like the people of the early 80s who said that RMS's campaign was utopian.

(As for Windows7Sins - success is judged on how much effort it cost to get the results it got. Maybe FSF found someone who'd do it really cheap. Maybe it cost next to nothing, and whatever result it had is a bonus. Not every campaign has to change the world in order to be a positive force.)

The xkcd guy has drawn RMS a few times. What makes you think that RMS's proselytising wasn't the inspiration for that comic you loved?

The Party of Gno (Linux Magazine)

allquixotic — Sat, 19 Jun 2010 21:45:58 +0000

I don't think Free Software is incompatible with SaaS or cloud computing / cloud storage. In the case of Ubuntu One, it makes perfect sense to release the software (all of it, not just the core) that runs the system. They already did it for Launchpad; they can (and should) do it for Ubuntu One. Here's why.

All software that is written to require Internet connectivity to a remote server is of a special class of goods, because its supply/demand model is much closer to the traditional model of selling physical goods. The *effective use of* the software requires a non-replicable physical good, where the production costs of each unit (one unit being a server computer residing in a datacenter, with physical connectivity to the Internet) are linear. So you can sell portions of the resources of your company's server assets running your cloud software. Giving away your software is not the same as giving away resources on your server. People will still use your service, because the alternative is for them to purchase their own, non-replicable physical assets needed to run the software (i.e., their own computing assets). Unless they have the same level of redundancy, professional IT technology, server-grade hardware, and best practices that your company maintains, they will be at a distinct disadvantage, putting their data at risk. Customers will recognize this -- even consumers -- when data starts getting lost due to disk failure, or outages start occurring because some idiot hit a telephone pole and knocked out power.

Sure, you will be allowing competitors to also buy physical hardware and undercut you using your own software; but the fact remains that they actually have to *do* that. They'll need a big investment, maybe venture capital, to get the required resources to compete with you at scale. You are then challenged to go to vendors like Intel, AMD, Nvidia and IBM to acquire hardware that is cheaper or more efficient (more power per watt; more resources per dollar) than your competitors, so you can lower your prices and still make a good profit.

And then there's the fact that, if their data is already stored in *your* system, and their friends are all connected to *your* system, and their client programs are configured to connect to *your* system, it's quite a bit of effort for someone to eject from their social, financial and time investment in your system, and migrate to someone else's. And since you wrote the software, you are necessarily the first business to offer the cloud-based service surrounding it, which means the first users have to use your system, due to lack of alternatives. If you want to keep someone from competing with you while you're still developing the software, just don't distribute your free software -- to anyone -- until you have your own service online and available to the general public, at which point you can then distribute the source.

This makes the SaaS / cloud computing market look like almost any other commercial physical goods service. But the benefit to the public of releasing the software that runs your system under a Free Software license is obvious to anyone who reads LWN.

Other than these points, I like the direction the original article points in. I am a card-carrying associate member of the FSF, but I agree with the author that moving towards positive campaigns is the way to go. I am still a bit undecided about the author's friendliness towards commercialization, but the general concept of moving away from negative campaigns in favor of positive campaigns is a novel concept to me, that I think would propel the FSF into a much more positive (and popular) public image.

Please give me fewer articles, and more diamonds

sebas — Sat, 19 Jun 2010 16:47:55 +0000

But those groups are only different in focus and mission, not necessarily in personnel. That is, many "FSF" people *also* write code.

It's really two-fold, you need to tell people why Free services matter, just like Free software does, but you also need to write code. The FSF is about the "tell" side, telling them to write code instead of talking about it is beyond the point.

History editing

bronson — Sat, 19 Jun 2010 16:24:10 +0000

Here's my oversimplified view:

Open Source: MIT, BSD, LGPL, or Apache licensed. The developer just wants to get her code in the open, see others use it, and start receiving patches. If someone else tries to close the code back down, that's unfortunate but it sure doesn't keep her up at night.

Free Software: GPLV2orLater, GPLv3, AGPLv3: The developer can recite the Four Freedoms, cares deeply about future developers, and is willing to expend a lot of effort to lock those freedoms in stone. If it means a license will be huge and complex, or GCC will not have a plugin interface because it could be misused by proprietary developers, then that's a price worth paying.

By this, I think Zonker is right. In my experience, and as evidenced from the astounding percentage of Apple laptops I've seen at every developer conference I've gone to in the past few years, the more pragmatic Open Source movement is thriving and the more idealistic Free Software movement is shrinking.

Personally, I know the FSF has done a really good job of pushing me toward the Open Source camp with GPLv3. I'd hate to use a license so complex for any code that I release! So maybe I'm not the most impartial party to be attempting this observation. :)

iPhone =/= Debian app

job — Sat, 19 Jun 2010 16:21:20 +0000

Hardly. Apple's appreciation of web standards is because they've been on the wrong side Internet Explorer, and that hurt.

Gno Thanks

mmcgrath — Sat, 19 Jun 2010 13:58:03 +0000

> I'm just saying that the advertising that Apple does, and the unpaid placements they get (like on The Colbert Report for example)... really helps create a want and drives sales.

So does having a highly functional operating system. If OSX was free, or the Linux desktop was paid only, we'd see an even more major shift to OSX then we do now. It's because we have all the people that care about freedom over function and unfortunately for us, it's a very tiny minority of users. The larger majority prefers function and the Linux desktop offerings just don't compare to the non-free experiences.

conflation?

wingo — Sat, 19 Jun 2010 13:12:01 +0000

From the quoted snippet:

And I'll freely admit, I've advocated the pragmatic approach because after more than 10 years of working in the community, it's clear that getting things done with a purist approach isnt working.

Not sure what you're getting at here, zonker; might you be conflating "negative" with "purist"? Surely there are ways to be an idealist and to be positive.

History editing

jzb — Sat, 19 Jun 2010 11:19:28 +0000

"Claiming that the FSF isn't doing enough is just absurd:"

They're not doing the right things. DBD is a complete and utter waste of time. Win7 Sins is a complete waste of time and FSF contributor dollars. RMS spending his time on the GNOME foundation list arguing for "GNU/Linux" instead of spending time constructively is a waste of his time and the time of other people on the lists...

"Is it not pertinent to ask where the Open Source Initiative stands on these matters, or do we hold some organisations to a higher standard than others?"

OSI, last I checked, wasn't seeking funds to squander to run pointless campaigns.

The Party of Gno (Linux Magazine)

jzb — Sat, 19 Jun 2010 11:15:31 +0000

Yeah, Bruce - sure. It's easy for you.

If the only folks you care about are other people like you, then ... great. Enjoy your freedom while the general public keeps using non-free software and becomes further entrenched in non-free software. I'd like to see free software for everybody, not just the people who happen to have the technical chops to maintain their own server + services.

Not to mention the financial ability to do so. Lots of folks can't afford $120 and up per year to rent a server of their own.

Please give me fewer articles, and more diamonds

jzb — Sat, 19 Jun 2010 11:12:08 +0000

I'm sure people said the same thing about UNIX. And databases. And desktop software...

It's funny that FSF-proponents are willing to try the impossible (convincing the general public to stop using things like Facebook because they're non-free) but unwilling to try the possible but fairly difficult (replacing said services with FaiF services).

I guess it is much easier to proselytize than to plan and program...

iPhone =/= Debian app

asaz989 — Sat, 19 Jun 2010 08:49:40 +0000

Spot-on article. On a side note, this is what I think is behind Apple's promotion of web standards - they don't think that the iPhone-app craze is sustainable, if only because people now have non-Apple smartphones and developers will not put in that much effort for a smaller bite of the mobile market. Apple wants the iPhone to still have consumer appeal if and when sites realize that web pages should be designed for web browsers (perhaps with a touch-screen version or stylesheet), and not for any specific platform.

iPhone =/= Debian app

k8to — Sat, 19 Jun 2010 08:11:52 +0000

Sure, but the developer paid several hundred developers this year, and so did you, to Apple. And both of you are going to pay several hundred dollars again, when the next revs comes out in another couple of years. That's a bigger deal.

LSB? (was: He's not getting it)

AndreE — Sat, 19 Jun 2010 07:41:34 +0000

Not to mention that the actual format of the package is actually not even the problem

We still have to consider naming schemes and separation guidelines, and other packaging issues (e.g. Fedora's non-bundled system libs).

Just getting everyone to play RPM doesn't help. Just Try installing a fedora rpm on a suse or mandriva box to find out.

LSB? (was: He's not getting it)

da4089 — Sat, 19 Jun 2010 04:45:55 +0000

LSB is a set of minimum guarantees for things an LSB-compliant system will provide.

But it's so minimal as to be useless. Last time I tried to package an LSB-compliant application, it didn't provide libxml, OpenSSL, zlib, etc.

So your application has to package all of that stuff internally.

And then, in 6 months (or 18 months for enterprise folks), the whole crop of sexy stuff (think of changes like Xorg, PulseAudio, DBus, etc) that's available for native packages isn't available to your LSB application.

It's a well-meaning, but busted model.

iPhone =/= Debian app

bartszyszka — Sat, 19 Jun 2010 04:40:00 +0000

I don't think your analogy to websites during the dot-com bubble works exactly. During the dot-com bubble, people gave everything away for free without a business model, running on borrowed time until the investors' money ran out. With iPhone apps, the actual customer pays for what the developer makes (with Apple taking a cut, they are a business afterall). I, as a consumer, pay $9.99 for an iPhone app and the developer gets a majority of that money and developers make real profits. That's a big deal. There are very few websites people are willing to pay even a dollar for.

The Party of Gno (Linux Magazine)

dps — Sat, 19 Jun 2010 01:46:51 +0000

I am not sure all of the companies that contribute to the free software pool ever where free software purists. The cheapness and flexibility of free software appeals to commercial users. My current work also appreciates the good 64 bit support.

You could push the flexibility, support for your target environment and impact on your cost base as reasons to prefer free software. Using free software avoids the need to track the royalties owed to lots people under dozens of different licenses, which is likely to be expensive.

You can also argue for contributing your changes back to the community on pure cost grounds.

I know my work, which writes commercial software that runs only on linux x86_64, values the qualities above and can find commercial reasons for feeding patches they require upstream. The GPL is not regarded as a problem.

Reinventing The Party of Gno

crosstan — Fri, 18 Jun 2010 22:51:46 +0000

I think you are trying to reinvent the wheel. We have any number of attempts at the party of Gno...

The party of Gnovell

The party of Sco.

All trying to compromise the principles of free software and doing such a good job (at compromising anyway).

Please give me fewer articles, and more diamonds

wmf — Fri, 18 Jun 2010 20:34:04 +0000

I wasn't joking, and I disagree with you again. I think at some point Google and Facebook will become so entrenched that there will be nothing us freedom fighters can do about them. GNU has made great progress on a Unix clone, but I'm not so optimistic about social networking; it looks like the proprietary walled gardens are improving faster than the open stuff so the gap is widening, not closing.

Gno Thanks

davide.del.vento — Fri, 18 Jun 2010 20:32:25 +0000

> The first GPLv3 drafts allowed a Affero-like provision
> to be added as an *optional* clause.

I know and that was for flexibility and compromise too. The no-compromise would have been GPLv3 with mandatory Affero provision, always on.

Please give me fewer articles, and more diamonds

wmf — Fri, 18 Jun 2010 20:28:15 +0000

Rewriting all the social Web 2.0 stuff that people want as AGPL (as Zonker proposes) will be expensive. Re-architecting all the social Web 2.0 stuff that people want into some encrypted P2P system powered by Guile will be really really expensive. It looks like you're biting off way too much IMO.

The Party of Gno (Linux Magazine)

cesarb — Fri, 18 Jun 2010 18:38:21 +0000

Even if it is not a barrier to purchase, all else being equal, DRM-free content has better fitness than DRM-restricted content.

For instance, while with DRM-restricted content you might have to carry several large physical round objects with you to wherever you might want to use the content, with DRM-free content you could just duplicate it everywhere you might want to use it, carry a copy of it with you within a few very small rectangular objects, and even get a quick copy from a friend if you forgot yours. DRM-free is clearly superior here, and it becomes even more superior when new uses which were not tought of by the DRM makers are found (suppose for instance you want to classify your movie collection by the average color of each movie; the DRM will probably not let you).

Not to mention that, in the particular case of free (as in freedom) content, the amount of available content tends to only rise. If I have $50, I can have all the free content ever created plus $50 of non-free content*. Given time, the amount of available free content (which is of course DRM-free) will be large enough that the annoyance of having to put up with DRM starts making less and less sense ("You mean I cannot watch this movie I just bought in my high-quality computer monitor, which is what I use to watch every single one of the other movies I have?").

The extra flexibility of DRM-free content means it gets used more, which means it brings more value to the user, who then will seek more DRM-free content (since after all it is better value). (But remember that this is "all else being equal".)

* Simplifying out the costs of network access, disk space, and so on. If you include them, it looks even more unbalanced; it is either $50 of non-free content or $50 to pay for enough bandwitdh and disk space for a lot more content, so you might decide not to buy the non-free content after all and use the money instead to pay for more bandwidth and disk space.

Open vs. Free

dbruce — Fri, 18 Jun 2010 18:02:18 +0000

AFAICT these terms refer to almost precisely the same bodies of software. Once you specify that "Free" means "Free as in Four Freedoms", and that "Open" means "Open Source Definition" and not just "the code is visible", Free Software and Open Source Software are practically identical.

Of course, these terms have very different implications with regard to the goals and beliefs of the developers. Nonetheless, it makes little sense to claim that the use of open source software is thriving, but free software is not. The author obviously knows this - I assume he is saying that the devs and users of Free/Open software seem to be caring less and less about freedom as the years pass.

The Party of Gno (Linux Magazine)

lmb — Fri, 18 Jun 2010 17:56:37 +0000

There's a huge difference between having DRM on your content or DRM on your software licenses, too.