Hacker News Reader

Comment by LeFantome | Hacker News Reader

I am happy to see Alma takes this path.

Now they can actually be a real distribution and a true member of a community. They can make contributions, fix bugs, and add extras if they choose.

Now they will be “based” on RHEL ( ABI compatible ) but not “identical” per se.

This is what Red Hat wanted to happen. Hopefully this puts an end to all the Red Hat has gone proprietary nonsense as well. Pretty much everything Red Hat does will go into Alma as well. Alma will be built on the back of Red Hat’s efforts but it will be in a proper community way and they will share code and not just directly rip off the whole distro.

View on HN

> not just directly rip off the whole distro.

I literally love this take, because it can't be more wrong.

Most of the CentOS, Alma and Rocky Linux users used these distributions, because they were able to find their way and fix their problems without asking RedHat's help, and RedHat was explicitly saying that they're selling the "support", not the distribution.

Most of the people managed these systems had the chops to rebuild everything from bottom up, if required, and report the bugs the correct places to support everyone, including Red Hat.

Red Hat changed its stance overnight and decided to sell the distribution instead of support, closed the SRPMS they provided, flipped the table and screamed "Thieves!".

It's funny and sad at the same time. They started this model, lived in a win-win relationship for two decades, and changed their minds overnight like a drunk sailor, because IBM wanted more monies.

This will be very harmful to everyone involved in this in the medium-long term, because RHEL will lose the mind share which supported them on the enterprise space.

IBM's software products may be locked to RHEL in the short term, and a RHEL license have to be bundled with their sales, but it'll be a niche, dark room distro, not a well known one.

We'll see how Fedora will fare from IBM "Community"'s bright ideas like "opt-out before the first send" telemetry.

Alma and Rocky directly competing with RHEL, undercutting RH's business by rebranding RHEL clones to sell cheaper support (since they don't suffer RHEL development costs) is what changed.

Why should RH help bad faith competitors?

Being a corp slog nearly my entire life, it's good to stay open to all opinions. With that I say, that I have encountered piles of fraud with organizations in my career. They would buy a couple licenses of RHEL when their fleet was really CentOS and they would just lie on support calls. I think this is what Red Hat was getting at with all the "freeloader" nonsense. Tone deaf, yes. Wrong, not totally.

...continuing to be fair, Red Hat support has literally been the best I have ever used in my career. Hands down. I know I sound like a shill but I'm not. Just a reg ol' engineer. I've since moved on, like most, to cloud and past RHEL but it has it's uses. Cattle not pets for cloud!

Regarding the "freeloaders" comment, Mike McGrath made a follow-up post on LinkedIn that includes the following:

> Finally, I wanted to say something about the term "freeloaders" I've seen many use it. This is a mostly internal term we have at Red Hat, it looks like at some point it slipped out in the public. So what does it mean? A freeloader is when a large enterprise business has 20 RHEL licenses, 150,000 community rebuild systems, and sometimes hundreds of user accounts and hundreds of kbase searches per month. It's not the enthusiasts, it's not the hackers and coders, it's not the academics, and it's not the people that use rebuilders because they can't afford it. We really try not to use the term, but when we do, it's about the large companies that can afford to pay but don't.

Sorry, no. Red Hat broke the spirit of the GPL. The social contract of FOSS is not that you get guaranteed a viable business model or maximum profit.

Red Hat went as far as they could without breaking the letter of the GPL. They likely would have gone full proprietary if they could have legally done so.

Whether you see a problem with that depends on your values. A new school of thought has arisen, let's call it COSS - capitalist open source software, in which the purpose of open source is for large companies to flood the market with a loss leader to kill off smaller competitors, and for individuals to pad their resumes.

Under COSS, open source is merely another tool to feed the capitalist machinery, so Red Hat's strategy seems only natural. But many of us do not share these values.

Isn’t GPL the HN’s favorite boogeyman? Boooooo, the license is not permissive enough. Boooo, is not true freedom. Boooo, whatever shall my poor startup with millions in funding do if it’s GPL. Booo, but what about all those lost exposure points because Amazon/Google won’t be using it.

But look how are the tables turned if it’s some other guy exercising his rights.

> Boooooo [...] booooo

Sorry, I fail to see how this addresses any of my points.

> But look how are the tables turned if it’s some other guy exercising his rights.

If we're talking about IBM/Red Hat here, they are not exercising their rights.

Can they close the SRPMs? Yes they can.

Can they tell "You can't redistribute this GPL licensed source code", haha, no. They can't even tell you to not redistribute BSD/MIT/Expat/WTFPL[0] code. They can't override these legal texts with EULAs.

GPL is working as intended, and irritating corporations the way it should, and yes, what Red Hat done is treason, not because they closed the SRPMs, but the collective things they did.

[0]: http://www.wtfpl.net/

Red Hat's customers still have the right to redistribute the GPL-licensed source code from RHEL, but Red Hat may end their business relationship with them if that right is exercised. This is still in line with what the GPL requires of Red Hat.

Assuming what you say is true, isn't that threatening the customer to not exercise their rights?

Isn't this limiting in a legal and practical sense?

How this is ethical, and allows unimpeded distribution of FOSS code?

Also, I still need to read and see the agreements myself.

> I still need to read and see the agreements myself.

See the 'termination' section of the enterprise agreements from the Red Hat licences page.[1] What constitutes a material breach is specified in Appendix I from the Product Appendices section, and it includes the redistribution of software obtained through the subscription (sections 1.2.f and 1.2.g of Appendix I).

> How this is ethical, and allows unimpeded distribution of FOSS code?

It is designed to discourage the distribution of the code while keeping Red Hat in compliance with the GPL. So it cannot prevent the customer from exercising their GPL rights, but a continued business relationship with the software vendor is not something that the GPL protects. PaX/grsecurity had used a similar tactic with their Linux kernel patches.

1. https://www.redhat.com/en/about/agreements

That's what most are concerned about. For a company that markets themselves as "open" or "the open source company", seeing their original massive OSS project now only achieve what's 'required' and nothing more... that doesn't seem in line with the expectation set over decades of more magnanimous behavior before.

> seeing their original massive OSS project now only achieve what's 'required' and nothing more

If they were only doing "what's required" they would have ditched their upstream first approach to fixes and the only ones seeing their code would be paying customers, or people who the paying customers distributed the code to. The fact that they are still one of the primary contributors not only to the upstream linux kernel but countless other open source projects pretty much by definition means that they are still going well above and beyond "what's required". If you don't like their direction, fine, but pretending like they are entirely freeloading and doing nothing for the broader open source community really weakens your position.

One thing I've been wondering--if a RH customer decides to surreptitiously distribute the source, how is RH going to know which customer did it? I guess they could do something like making each copy of the source unique by slightly varying the non-string white space in the code.

I guess the other issue would be that distributions like Alma or Rocky (and, to varying extents, their users) probably value stability and reliability too highly to be comfortable relying on such a janky form of source distribution.

Reply on news.ycombinator.com