User: Password:
|
|
Subscribe / Log in / New account

glibc got forked

glibc got forked

Posted Jun 27, 2013 11:27 UTC (Thu) by dakas (guest, #88146)
In reply to: glibc got forked by Wol
Parent article: Richard Stallman becomes a Hall-of-Famer

When one is catering to a wide range of systems loosely held together with a mixture of strongly and weakly followed standards and de-facto standards, "is correct" is pretty much tantamount to a reasonably reliable outcome of "it works". If the goal is "it works with Linux, let all the rest be damned", of course things are easier. And if you add "don't bother with upstream, just fork", things are easier still.

At first. But this kind of "everyone out for his own" was not the idea of Free Software. It _was_ more or the less the starting position of Linux. But you'll find that nowadays most of the more successful applications on Linux are portable to other platforms and are maintained in a manner that is not focusing on only one operating system, and that are having developers from several different backgrounds and platforms.

And indeed, software that is radically catering to just a single platform tends to become uncompilable even as that single platform develops further: a later version of some platform will tend to change in aspects that are not standardized and/or common with other platforms.

I have some code here that I adapted to SVGAlib once since that was easier to do than to adapt to X11. Net result: it won't even run under Linux-based systems nowadays.

It was not just libc: even the file utilities were forked at some point of time, and the trend was "let's not cooperate with upstream, that's too cumbersome and GNU is uncool".

I'd call the "GNU/Linux" campaign silly if it were not for the fact that it was an effective part of the work for getting people care about working as a community centered around sharing software beyond one's own immediate needs.

And indeed, nowadays we are sadly in the "but it works!" situation regarding the Internet and its use for communication. Freedom and privacy take a second place towards raw function, and the spy organizations are having a field day. The situation is much worse than it ever was in Stasi-riddled East Germany, and much worse than what Nixon got impeached for. There is no reasonable accountability of security agencies even to their own government, let alone their people, any more. And the omnipresent excuse is "it works, so let's not look too closely".

Which is what has given the U.S. a basically uncontrolled NSA (all control mechanisms are internal and optional) and lackadaisical or flamboyant stances towards war, assassination, torture, confinement without due process, wiretapping, unwarranted arrest and so forth and so on.

It's obvious that in the case of the Internet it is again left to activists to create turnkey endpoint encryption infrastructure that will create a haven of privacy for the selected few using it, like Free Software created a haven of software freedom for the selected few using it, and work towards its ubiquity.

And if that is supposed to eventually take hold and provide benefits for more than just the people that really care, one needs to look out for the interests of more than just oneself.

It is a pity that governments don't understand that they are the servants rather than the master of people. But while that is the case, one needs to work within the given frameworks to create colloboration and freedom.


(Log in to post comments)

glibc got forked

Posted Jun 27, 2013 11:45 UTC (Thu) by mpr22 (subscriber, #60784) [Link]

In the 1990s, you had to sign physical legal paperwork and mail it to the FSF to contribute code to GNU, while you could contribute code to Linux, or to Linux-world forks of GNU, just by writing code and handing it over. It is not surprising, and not particularly blameworthy, that the Linux world forked GNU.

glibc got forked

Posted Jun 27, 2013 14:12 UTC (Thu) by dakas (guest, #88146) [Link]

Just for the record: for "key components" of GNU, like Emacs, GCC, and very likely glibc, you still have to sign physical legal paperwork to contribute to the GNU-maintained versions of software. You make it sound like this practice ceased at some time after the 1990s. It is still very much in place.

One consequence is that this kind of software could be made GPLv3+ without having to get back to every contributor. While this would also have been possible with the usual GPLv2+ licensing, other licensing decisions (like the special lenient licensing for Bison-generated code, or for some static library stubs, or the division line between GFDL-licensed documentation and GPLed code) really require the copyright holder to act.

It's also telling that the SCO/IBM copyright case disaster (that is still dragging on) happened over the "just by writing code and handing it over" managed Linux kernel rather than any of FSF-maintained software.

Now it turns out that IBM is quite likely to prevail in this case, but SCO will go out of business without having any substance left to pay damages, court costs or whatever else to IBM, leaving them with accumulated costs likely going into hundreds of millions.

If that kind of court case had landed at the doorsteps of the FSF rather than IBM, the FSF would long be broke now, never mind that they would have been in the right.

So this is again a case where the choices of the FSF have proven to be a nuisance quite above the "let's just do what works and nothing else" stance, but a nuisance that seems to have been prudent in hindsight.

Again, it is a sad statement about the world how many of those overcautious and paranoid decisions have proven to be quite justified in hindsight.

glibc got forked

Posted Jun 27, 2013 15:22 UTC (Thu) by mpr22 (subscriber, #60784) [Link]

I find SCO vs. IBM far more telling of the easily-exploited nature of the USA's (a) court system and (b) bankruptcy protection laws, than of anything related to what is a good contribution policy. IBM clearly thought it had a sound position when contributing the code in question to the Linux kernel; SCO acted as if it thought differently, and was able to spew sufficiently convincing legal bafflegab to persuade a (possibly overly-sympathetic) judge that their case was not completely devoid of merit. It's not clear what Linus having a copyright assignment requirement would have done to avert SCO v. IBM, unless it prevented IBM from ever participating in the first place.

glibc got forked

Posted Jun 27, 2013 16:10 UTC (Thu) by dakas (guest, #88146) [Link]

The main part of the court debacle were the claims of code getting copied over from UNIX into Linux. This has not actually been resolved, instead Novell prevailed in court (in a separate case) with its claim of never having actually transfered copyright to SCO. Now if any successor in interest of Novell would want to restart where SCO has left off, that would be feasible. Except that SCO's case had already been looking quite loco even assuming that they had copyright of UNIX: some of the things they complained about were provably actively contributed by Caldera employees, some stuff had just been lifted without attribution from BSD and they complained that they found it in Linux and so on and so on.

Nevertheless, SCO's opponents have paid a _lot_ of money to defend against this nonsense. The FSF has never been drawn into anything even remotely as extensive as that, and part of the reason may well be its assignment and attribution policy.

Whether or not you want to assign the blame for this fiasco to the U.S.A. court system, it is the court system that the FSF has to work with.

glibc got forked

Posted Jun 27, 2013 16:52 UTC (Thu) by dlang (subscriber, #313) [Link]

given that SCO sued IBM about code that IBM wrote and contributed to the kernel, I don't think that there is any possible assignment scheme that would have prevented the SCO lawsuit.

SCO was ruled against multiple times, but appeals courts have given them second and third chances, but even so, 99.9999% of what they originally claimed has been thrown out, even without the Novel intervention.

At this point, if Novel were to try and pick up where SCO left off, they would not get very far. Between all the work that went into showing the weaknesses in SCOs claims, and the estopple that would be able to be claimed by anyone sued, there is basically no chance of anyone winning.

Now, that doesn't mean that nobody could take a case to court and eat up a lot of legal fees on the part of the defenders, that doesn't take having a great case, it just takes having a great lawyer who can convince the Judge that they may have a case.

glibc got forked

Posted Jun 27, 2013 20:12 UTC (Thu) by anselm (subscriber, #2796) [Link]

The main part of the court debacle were the claims of code getting copied over from UNIX into Linux. This has not actually been resolved

A necessary prerequisite for the »resolution« of that claim would have been for SCO to come up with some (non-trivial) code that had actually been copied – by IBM – from (proprietary) UNIX to Linux. Despite loads of public noise to the contrary, and despite being ordered multiple times by the judge to produce something – anything, really – to support their claim, SCO couldn't present anything interesting in court, so the issue became moot.

glibc got forked

Posted Jun 28, 2013 7:00 UTC (Fri) by micka (subscriber, #38720) [Link]

> The FSF has never been drawn into anything even remotely as extensive as that,
> and part of the reason may well be its assignment and attribution policy.

I don't think so. I think it's more sensible to assume that it's because the FSF doesn't have the financial resources of IBM that it didn't happen.

If you're SCO and want to get a massive amount of money by trolling some other entity, you don't target a company that doesn't own a massive amount of money.

glibc got forked

Posted Jun 28, 2013 10:13 UTC (Fri) by anselm (subscriber, #2796) [Link]

You also need to own (or believe you own, anyway) something that the other party can plausibly be accused of having ripped off.

On its face the claim that Linux, being very much like Unix and having been worked on by companies like IBM which also worked on Unix, might include at least some proprietary Unix code is not completely out of this world – at least to people who are not well versed with the history of Linux. On the other hand, the whole point of the FSF was to develop a »workalike« system from scratch in order to give it away, so the FSF would be a much more difficult target to attack in the first place, on top of their not being quite as rich as IBM.

(Some people thought at the time that The SCO Group was basically betting the farm on making themselves obnoxious enough to IBM that IBM would acquire them just to shut them up and make the lawsuit go away, which turned out to be a fatal miscalculation.)

glibc got forked

Posted Jun 28, 2013 2:17 UTC (Fri) by jschrod (subscriber, #1646) [Link]

> It's also telling that the SCO/IBM copyright case disaster (that is still
> dragging on) happened over the "just by writing code and handing it over"
> managed Linux kernel rather than any of FSF-maintained software.

This statement is so dishonest, it's mind-boggling.

As if FSF copyright assignments, or any other scrutiny of code submissions, would have stopped SCO, or influenced their legal action in any way. That it is still dragging on, even though court has decided *that they don't have any copyright in the first place* is enough to show that.

Could you, as a guest, please go back to some other site, where you come from, and troll there?

GNU contribution hassles

Posted Jun 29, 2013 22:28 UTC (Sat) by giraffedata (subscriber, #1954) [Link]

It's also telling that the SCO/IBM copyright case disaster (that is still dragging on) happened over the "just by writing code and handing it over" managed Linux kernel rather than any of FSF-maintained software.
This statement is so dishonest, it's mind-boggling. Could you, as a guest, please go back to some other site, where you come from, and troll there?

It sounds like an honest mistake to me. Cases like this usually involve someone accidentally contributing someone else's code. Like an employee does it without the employer's knowledge because the employee doesn't know creates liability. More formal contribution regimens prevent that.

That wasn't an element of the SCO case, but the case was fairly complex, and I wouldn't hold it against someone for misremembering it that way.


Copyright © 2017, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds