I think his view that the most important part of the license is that it makes the software better is shortsighted. If linux gets used in 95% of all capable hardware everywhere, but it's all on locked down hardware by a few companies, what is the point of being open source? Effectively, it's proprietary software which is co-owned by a group of corporations, and the open source is primarily a training program for future employees and phds. Users being able to choose what software they want to run on their devices does matter. Well, he did say it was a fine license (if you care about that). Also, I don't think he gives enough credit to fsf, as almost all other free/open source licenses automatically and always allow later versions, but gpl doesn't. Creative commons is one big example.
He doesn't have a problem with the GPLv2 allowing later versions, that's completely reasonable. His problem was that with the GPLv3 they changed the spirit of the license. He thinks a license "upgrade" shouldn't have done that, they should have just released a new license instead.
I.e. previously the agreement was "here's my code, if you change it send me the changes", now in addition to that it includes clauses about what you can do with the hardware you run it on.
I also don't want a future where all you can buy are locked-down devices, but is the GPL really the right venue to fight that battle? If companies want locked-down hardware they're just going to do lock it down anyway, and just use e.g. FreeBSD instead of Linux and contribute none of their code changes back.
Isn't it better that they release locked-down hardware but contribute their code changes, so that if someone wants to make competing open hardware they can start with all the same driver & feature support in Linux?
I believe the spirit of the license was not changed. The spirit was spelled out by fsf and rms very well in things like the free software definition, and faqs about the purpose of the license etc. Not disallowing tivoization was an oversight in the legal code which needed to be corrected for the spirit to remain what it had been. Linus thinks the spirit was in the exact legal code. That is more like "the letter of the license", not the spirit.
> I also don't want a future where all you can buy are locked-down devices, but is the GPL really the right venue to fight that battle?
Is it legal for software to defraud people, to act explicitly for ransom/extortion, or purposely kill people by stopping their medical devices? No. If it was, and it was happening a lot, I bet a copyright license that had something to say about that would be popular. Consumer owned devices which attempt to force people to use specific software for no technical reason, just so they can make more profit from the user, should be illegal outside of any copyright license. Until that happens, then yes gpl is one of the right venues.
> Isn't it better that they release locked-down hardware but contribute their code changes, so that if someone wants to make competing open hardware they can start with all the same driver & feature support in Linux?
Ya, right now it's that way, especially since they don't do a good job of locking things down and you can circumvent it on most phones. But it's not at all certain it will stay that way.
It helps if you read the license before talking about the spirit of the license. In the very first paragraph it says: "the GNU General Public License is intended to guarantee your freedom to share and change free software".
If I can't change the software that runs in the device, then I can't change the software. The letter and spirit of the license, ie, to use, share and change software, is fundamental broken if you add restrictions to use, share or change the software.
To take an other quote: "you must give the recipients all the rights that you have". Again, letter and spirit is stated directly. If you have rights to use, share, or change things, give that too to the recipient.
(all quotes taken from GPLv2 license text - Preamble).
I am really heartened by his push for consumer-facing simplicity of the OS which, by his account, the distros have not helped in achieving. I didn't expect that from him given that he operates at a much fundamental level of computing. But his example with his kids' connectivity issues tells me that even the nerdiest person starts appreciating the value of simplicity when they start facing use cases that demand it.
Loved his rant on glibc, etc. breaking stuff. I've been saying this myself for years. Really static linking is the only way to make a binary for Linux that will work on any distribution.
It's interesting that a lot of this isn't a problem in FreeBSD due to the Base system and ABI guarantees. FreeBSD Ports also averts many of the grumblings about distro packaging.
Having worked with FreeBSD before in a Real Product, I'd have to say that the main reason FreeBSD doesn't have a problem with this is because most of it hasn't changed since the 1980s. Not even because it's particularly good, but just because nobody wants to work on large tracts of the codebase. The normal feeling is "bwah, this is complicated and ugly. I'll just leave it alone..."
I'm not sure, but doesn't the opensuse build service tries to solve these problems? I haven't used it, but I think i slurps source code and builds packages for multiple distributions.
And I absolutely agree with Linus view on linking. I just tried to install ffmpeg on opensuse on a desktop pc. All I wanted was one statically linked binary. I had to dig a wiki, follow outdated build instructions (not nearly enough dependencies where listed, I had to go back several times, configure runs like 20 minutes), it took hours. For one simple binary. FFmpeg could put on their homepage.
Build services are great if you have the source to everything. But the reality is I've got software I paid for on Linux that doesn't run on a modern distro after a decade, and software I paid for on Windows that's 15 years old and runs fine.
That doesn't solve the problem for commercial software and only works around the problem for open source.
In fact I ran in to this problem today. We build on CentOS 5 in order to support some customers who still use it. Unfortunately we linked against libssl.so.6 which is no longer available on newer distributions. We can't even statically link due to licence issues.
The man at 18.00 minutes with the question about the mailing list tone. He kind of called Mr. Torvalds out on the way he talks to people on the mailing list. What grounds does he have to say something like that to Torvalds?
Why not? First, I thought the response was perfectly reasonable, but second, I think the way Linus behaves publicly entitles any person with or without credentials to legitimately question if that is the best approach. Linus, by making his criticisms both public and humiliating at times must expect to be criticized by anyone, especially someone willing to come out from behind a keyboard and ask him in person!
The fact that Linus didn't take offense to the criticism of his abrasive nature shows that he expects it to be a two-way street. He really does have a thick skin.
Linus told a story about a guy, that wrote something for the kernel and
Linus thought in the first place that it was more or less bullshit, but
didn't told it directly to the guy, so when the guy was finished and
wanted to get his stuff merged, Linus had to tell him that he will never
merge this.
I don't know how mucher later, but this guy killed himself afterwards.
Linus told that he doesn't feel guilty about this or thinks that he was
the main cause for the killing, but well, I don't think that such an
experience has no effect on your behaviour, but I can't tell if Linus'
behaviour was different before this event.
I think that Linus told this story in the talk where he showed nvidia the finger.
Is it really so hard to believe that a man who is abrasive has no problem dealing with other people who are abrasive? There do exist different cultural attitudes and standards for abrasiveness, tact, manners, and respect. The standards currently used in corporate America are not universal; some people have stronger standards, others have looser.
> Is it really so hard to believe that a man who is abrasive has no problem dealing with other people who are abrasive?
That would be a classic symptom of narcissism. I'm not saying Linus is a narcissist, if anything I'm saying the opposite -- if he accepts the same treatment from others that he puts out, that argues against narcissism.
Now Steve Jobs, on the other hand ... I couldn't stand working with Steve because of his obvious and unbearable narcissism. He is why I never worked directly at Apple for more than a few weeks during many years of professional association on a number of software projects.
