[red_admiral]

There is free as in apache/BSD/MIT, and free as in GPL. In many parts of industry, going the "purist" route and putting your library under GPL is a sure way to make sure no-one will use it (the LGPL doesn't completely solve this problem either). Whereas more pragmatic approaches to free software have absolutely taken off.

Big names like React, Angular etc. are MIT licensed, for example.

It's like the FSF invented "free 1.0", and then others came along with "free 2.0" and it's a genuine improvement. The difference here is very much one that strikes at the FSF's core message.

[ta1243]

Being empowered to remove freedom is not a genuine improvement.

There are certainly issues with GPL software, either from the "you can't run modified software on your hardware" or "saas isn't distribution" get-out, or from the "small company X develop free software and charge for support/hosting/etc, so large company Y take that free software and charge for support/hosting/etc without giving back"

The former has attempts to tackle things with gplv3 and agpl, but the latter is a problem. A bunch of people make something cool like elsastic, or terraform, or which is free software, funded from hosting and support, then large beomouth with a near-monopoly on computing offers it with a different label as part of their provision. Companies prefer to buy from amazon/google/microsoft than random small company, and that seems somewhat wrong, but I don't see the FSF approach to marrying

1) how to keep software free 2) how to keep developers fed

When you have parasitical cloud companies

[globular-toast]

It's more like "free 0.5", not "free 2.0". MIT and other permissive licences are basically "free more me, but maybe not for you". There is only one possible reason that you want to reserve the right to restrict the freedom of others: you plan to restrict the freedom of others.

Individual freedom is easy. The King has always been free to do as he wishes. It's freedom for the entire community now and forever that matters.

[severus_snape]

Let's see 20 years from now which of React, Angular and Linux is still around.

[pjmlp]

After Linus and other key figures in Linux are gone, lets see how long it takes to still be relevant, or completly integrated in other big corp products, borg style.

Not that many of us can then watch it fold, as we belong to the same generation, however that won't change the evolution of computing world.

[SSLy]

With the non-enforcement of the license (see vmware, various chinese companies), Linux kernel is de facto MIT, yes.

[severus_snape]

My experience as an embedded software programmer is quite the opposite. We do care about releasing our Linux modifications under GPLv2, and we do contribute to community when it makes sense.

> see vmware

from https://lwn.net/Articles/696936/: "The court ruled that he had not provided enough specificity about the code he was claiming had been used by the company. The merits of the GPL and whether the two main parts of VMware's product constitute a derived work of the kernel were not even considered."

That doesn't seem to prove anything.

> various chinese companies

HiSilicon (https://github.com/hisilicon/linux-hisi) releases its fork.

[klardotsh]

And then as a counter-example, Boox frequently (always?) refuses to release kernel sources because... basically "we don't feel like it and nobody's ever going to be able to force us anyway" reasons, last I read. The latest development I heard in this space was that folks were trying to sic some YouTuber on them who has connections in China (I believe lives there, too) to try and guilt them into abiding by the copyright law they agreed to by using Linux.

It's... not a good time.

[cdelsolar]

React?

[orthoxerox]

And now we're moving to "free 3.0" (aka screw you Amazon), which is more restrictive than MIT or GPL.

[pipo234]

There are a lot of issues with FSF's perspective and the way they were written down. Running GNU derivative in a proprietary cloud is one of them. RedHat (in)ability to make a fair profit is another. Still Google's Linux "theft" led to more openness in Android than Apple's BSD Unix rip off.

Regardless of the license chosen, sustainability of open source is a problem. (I'm still baffled by the absence of government sponsoring; why do governments around the globe continue to spend billions on proprietary IT infrastructure, databases, and desktop computers?!)

Not sure if yet another license is the answer, but yes we could do with a new model of what "free" and openness entails.

[cxr]

> I'm still baffled by the absence of government sponsoring; why do governments around the globe continue to spend billions on proprietary IT infrastructure, databases, and desktop computers?

It's not a project that's talked about very much because it's not the trendy flavor of OSS dev that people generally have in mind when they refer to sustainability problems—which I suspect for most people means funding their favorite niche devops packages that are of dubious merit to begin with (and that we might very well be better off for not having in the world)—but it is pretty incredible that there isn't e.g. any well-known effort by at least one US government agency to put a fraction of what they pour into Microsoft into ReactOS instead, especially after the Windows XP crisis—and even if only to the extent that it should be made stable enough for that org's own narrow use e.g. with a particular piece of equipment/application. I never hear of anything like this. Not even a hint of trials or experimentation. Why?

This is the difference between GNU and the modern, GitHub-centric open source movement. For all the criticism about barriers to entry for non-technical audiences that have been leveled at GNU and GNU-adjacent projects, GNU has always been focused on users and the types of things that humans could conceivably actually want to use, and not the sort of devops shovelware that gets chucked into a repo and inexplicably chugs along with much fanfare because it's used by several dozen enterprises on the backend despite being of no real interest to an ordinary person. GNU userspace directly empowers millions of people (including many of those working on the uninteresting flash-in-the-pan shovelware) to do their computing (for work or for play) every day.

[callalex]

> I'm still baffled by the absence of government sponsoring

On the one hand, I agree, but on the other hand I am relieved that I don’t have to deal with people calling my OS “ObamaOS” and unleashing a barrage of hateful spam towards maintainers because of misguided political nonsense.

[rightbyte]

Redhat had been getting a profit for like 10+ years by now. How would that not be enough?

[ThrowawayR2]

Profitable, sure. But businesses that are poised for future success don't sell themselves, particularly to a has-been like IBM.

[rightbyte]

What would the stockowners care who pays them? I don't think they have much emotional attachment to Redhat.

IBM just have to pay more than the stockholders believe the stock is worth. It has nothing to do with if the owners believe in the company or not.

[froh]

what do you mean by "fair profit"?

[pipo234]

In case of Red Hat? Given their substantial contributions to Linux, primarily funded by support contracts, certification and consultancy I think they deserve to be profitable. At the same time, I as a user, also prefer to not pay them and have free access to their software.

For some time this was acceptable to all parties. But then companies not competing in Linux contribution started competing in the support contracts, certification and consultancy business, by giving away RedHat's free beer.

Wouldn't it be great if Oracle, Amazon (and Rocky, Alma) could continue to rely on RedHat's efforts but share / contribute / upstream revenue profits made consulting, certifying, supporting rather than freeloading on commons like RedHat? Forcing a distro fork the way RedHat IBM seems to want seems like an unwanted technical solution to a mere business/money problem.

[pjmlp]

MIT predates GNU.

GNU only took off due to AT&T wanting back the money they weren't allowed to charge for UNIX during its early lifetime, thus BSD uncertain future.

[severus_snape]

> and then others came along

GPL came after MIT (https://nitter.net/humphd/status/1112747178685026304)

> and it's a genuine improvement

How?

[red_admiral]

Let's say you want to develop software as a job, and make enough money to live. You also have the ethics not to go into the ad-supported track-everything mobile apps business. There is still good money to be made without selling your soul, but most of it is in the B2B not the B2C market. And there, you negotiate the terms of who owns what individually for each contract anyway, at least if you have any sense.

The problem with the GPL is that its "infectious" nature does not play nicely with how the B2B world works, where you might link against your client's in-house proprietary libraries, which even though they're never planning to distribute them, they very much don't want their own IP to become GPLed as a result.

So if you want to bring in a library to make pretty graphs or something, it very much matters what the licence is - MIT is generally fine though.

(How and whether a FOSS library author should be paid for the fact that one corporation used their library to develop software for another corporation, is an entirely separate matter.)

[teddyh]

> The problem with the GPL is that its "infectious" nature does not play nicely with how the B2B world works, where you might link against your client's in-house proprietary libraries, which even though they're never planning to distribute them, they very much don't want their own IP to become GPLed as a result.

Standard GPL FUD. Firstly, linking against proprietary libraries is fine as long as you don’t distribute the result to any third party. Secondly, no IP can “become GPLed” unless anybody chooses to make it so.

[red_admiral]

Some of the companies I've worked with, and their lawyers especially, have explicitly said that for them anything GPL in the application layer was a no-no (Linux as OS was fine though). This is on code that will never be distributed outside internal systems.

[teddyh]

Company lawyers being weird and wrong about GPL, film at 11. The GPL paranoia strikes deep.

[cies]

AGPL though is avoided by the plague by corp lawyers, and for good reasons. Hence I like it for some projects.

[type0]

> anything GPL in the application layer was a no-no (Linux as OS was fine though).

well, just proves how hypocritical it is

[zelphirkalt]

Copy-left licenses can work better for companies than non-copy-left licenses, if those companies are actually interested in improving the thing or creating the thing. They can release as GPL or AGPL and then profit from others contributing. If they released as MIT or other non-copy-left license, any competitor can use and modify without contributing back, which is the actual scenario any company should be afraid of. Copy-left actually helps companies that are not simply leeching.

[bonzini]

Google uses GPL software all the time and links it to proprietary software all the time. Since it runs on their servers and is never distributed, this is perfectly fine.

[goodpoint]

> don't want their own IP to become GPLed as a result

This is not how copyright works. No license can magically turn other code into GPL.

You can choose between using GPL libraries in internal proprietary products, or use LGPL libraries in published products or release your product as FOSS.

The choice is yours. There is no "infection".

[throw0101b]

'The Origin of the “MIT License”':

* https://ieeexplore.ieee.org/document/9263265

* https://news.ycombinator.com/item?id=34636092