CockroachDB relicensed [LWN.net]

The CockroachDB database management system has been relicensed; the new license is non-free. "CockroachDB users can scale CockroachDB to any number of nodes. They can use CockroachDB or embed it in their applications (whether they ship those applications to customers or run them as a service). They can even run it as a service internally. The one and only thing that you cannot do is offer a commercial version of CockroachDB as a service without buying a license."

to post comments

CockroachDB relicensed

Posted Jun 4, 2019 21:28 UTC (Tue) by jberkus (guest, #55561) [Link] (55 responses)

*sigh*. And I liked CDB.

Are ANY of the new databases still open source at this point?

CockroachDB relicensed

Posted Jun 4, 2019 21:36 UTC (Tue) by rahulsundaram (subscriber, #21946) [Link] (1 responses)

Maybe Postgres can become "new" and adopt extensions that make some of this possible?

CDB licensing model seems to mirror the original ghostscript licensing aka "Aladdin Free Public License" and GNU ghostscript GPL variant. I like that better than what Mongodb has done.

CockroachDB relicensed

Posted Jun 5, 2019 20:21 UTC (Wed) by k8to (guest, #15413) [Link]

PostgreSQL has a very good track record of eventually adopting new technology and integrating it fairly effectively with other functionality.

That said, a distributed SQL query planner for the level of spread CockroachDB aims for is a pretty specialized beast. I'm not sure that we'll see something like that in PostgreSQL for more than a decade.

CockroachDB relicensed

Posted Jun 4, 2019 21:55 UTC (Tue) by nilsmeyer (guest, #122604) [Link] (47 responses)

Do you want to offer it as a service for money or is it more about the license being a certain level of free?

Clickhouse fits the bill of being new (meaning from the latter part of this decade), a database and as well having a license that is considerably free (Apache License), however I think development is still mostly being done by a single corporation. An important difference is that the database isn't their main product.

I think we may see a lot of VC funded free software choosing a different license or altogether disappearing since their business model is extremely vulnerable to capture by the proprietary cloud vendors.

CockroachDB relicensed

Posted Jun 4, 2019 22:16 UTC (Tue) by SEJeff (guest, #51588) [Link] (46 responses)

And it seems like the takeaway of this (and the dozen or so before it, kafka and mongo included):

Open source software and venture capital funding don't go well together, especially when cloud vendors can take your software and compete against you.

CockroachDB relicensed

Posted Jun 4, 2019 22:22 UTC (Tue) by smoogen (subscriber, #97) [Link] (3 responses)

At this point in the dotcom boom/bust cycle.. many of the opensource VC backed companies were doing this in order to be bought so that their VC's could get some return on investment. I am expecting we are in the same "Step 3 of the Gnome Underpants Conspiracy" as before

CockroachDB relicensed

Posted Jun 5, 2019 15:28 UTC (Wed) by SEJeff (guest, #51588) [Link] (2 responses)

Having worked with you in the PHX2 datacenter to rack a server for the GNOME Foundation years ago, I thought there was some scandal in the GNOME Foundation I'd missed. Then I realized it was a southpark reference.

Well played sir. I think you're right fwiw.

CockroachDB relicensed

Posted Jun 5, 2019 18:37 UTC (Wed) by JoeBuck (subscriber, #2330) [Link]

When Eazel (the company that did the original Nautilus) folded their announcement included a link to audio from that South Park episode. Since I hadn't seen that episode, at first I thought it was a near-perfect spoof of South Park made to make fun of Eazel's business model, instead of actual South Park.

CockroachDB relicensed

Posted Jun 5, 2019 19:45 UTC (Wed) by smoogen (subscriber, #97) [Link]

Ha. That server is still there.. still doing things. I think

CockroachDB relicensed

Posted Jun 4, 2019 22:57 UTC (Tue) by timrichardson (subscriber, #72836) [Link] (38 responses)

Why do these firms choose open source licences to begin with, if their monetisation model requires them to restrict use of the code? It looks like they wanted the "sheep's clothing" of open source to build trust, gain contributors and gain market share. They call AWS the "bad guys" when all AWS has done is comply with the licence. The conflict is not between open source and AWS, it is a conflict between open source and a business model which requires exclusive control of the code. I feel sorry for contributors who have basically been mislead: open source is a community of trust, to avoid the "investing tenant" dilemma, where the tenant spends money to improve their property, only to have the landlord increase the rent since the value of the property has increased.

CockroachDB relicensed

Posted Jun 4, 2019 23:45 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

> It looks like they wanted the "sheep's clothing" of open source to build trust, gain contributors and gain market share.
External contributions for projects like this database are basically negligible. Typically less than 1% of the required work.

OpenSourcing was usually done in the hopes that selling a hosted version of the software is going to pay for development. OpenCore model is another possibility.

Unfortunately, both models don't really work.

CockroachDB relicensed

Posted Jun 5, 2019 19:44 UTC (Wed) by nim-nim (subscriber, #34454) [Link]

And, external contributions don't happen just by dumping some code with an open source license on github. Or by writing "open source" in all your marketing documents.

Contributions happen if the project is generally useful, split up in clean easy to understand modules, it its dep graph is sane and not full of obsolete vendored code no one wants to touch, if the buildsystem is simple to use and not a giant monolithic makefile, if you manage to build trust so others are ready to invest the time to write patches, knowing they have a chance to be integrated, and so on.

In other words, quite a lot of investment in time and energy, that distinguish entities which are serious about the open sourcing part, and entities that just want to greenwash a software created the proprietary way.

BTW, making a project generally useful, also means giving up on any exclusivity on the generated revenues. VCs hate that intensely, they will ask to make a product open source because it's a selling point, while sabotaging any actual open source activity, because they feel that's stealing from them.

CockroachDB relicensed

Posted Jun 5, 2019 0:17 UTC (Wed) by nilsmeyer (guest, #122604) [Link] (7 responses)

> They call AWS the "bad guys" when all AWS has done is comply with the licence.

That shouldn't be your measure of corporate responsibility. I see a lot of danger here when more and more software is moving into that walled garden. The end result may be that development stops and Amazon then just phases out the service in favour of their fully proprietary options. How is that good for anyone other than Amazon?

I have less of an ethical dilemma with those licenses as I have with the way Amazon is killing off these companies.

> The conflict is not between open source and AWS, it is a conflict between open source and a business model which requires exclusive control of the code.

I think the problem goes a lot deeper than that. As long as there isn't a viable business model for free software development the huge trend towards concentration on proprietary platforms is extremely worrying. Let's say at a certain point cloud providers start phasing out their classic, rented VM compute options or make them increasingly less attractive, how will you run your free software then?

Where are the successful open source business models?

CockroachDB relicensed

Posted Jun 5, 2019 3:14 UTC (Wed) by timrichardson (subscriber, #72836) [Link] (4 responses)

If you are open source, you can't charge someone for running your code on their CPU. So there can't be a business model based on that. It is strange to be surprised at this. Just as it won't be so nice to use Uber when the investor subsidy disappears. Open source changed the world long before it became the subject of speculative investments in a portfolio of startups hoping that one of them becomes a unicorn. I don't think we mourn the death of open source in this licence change, we mourn the loss of unsustainable goodies.

You could make money from training, consulting, support, bespoke customisations. But if your business model is "Let us manage your database, so you can focus on your business." (copy and paste from Cockroach Labs), then you are competing on managing a server. You are competing with the core business of AWS, and customers will probably find AWS a better choice. It didn't work out, so now there is a licence change which charges rent to AWS (or increases the cost of hosting Cockroach DB via AWS, so that the Cockroach Labs offer looks relatively better than now, which is the logic of tariffs).

Open source is not charity though, it is a contract of mutual obligation. Should we feel sympathy for Cockroach DB? What service have they provided open source? It is not hard to imagine a business choosing to use Cockroach DB because it was easy and cheap to run on AWS. They accepted the rights and obligations of open source. Imagine they found a performance bottleneck, and contributed code to fix it. In return, they received benefits of other contributors. This is a daily scenario for open source, and it shows the rational self interest of open source; it is not charity, it does not depend on "corporate responsibility". It depends on trust in the contract of open source. Now, however, they face being charged to use their own improvements, either directly as AWS passes on its new costs, or indirectly as they move to a less attractive service configuration.

Meanwhile, note the divide and conquer approach of the new licence, probably generous enough to distract many potential contributors to a genuine open source competitor. Perhaps we should ask if these 'look but don't touch' licences are a mutation of extend, embrance and extinguish? Ok, that's over the top, but I am far from convinced that AWS is the bad actor here.

CockroachDB relicensed

Posted Jun 5, 2019 5:44 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

> You could make money from training, consulting, support, bespoke customisations. But if your business model is "Let us manage your database, so you can focus on your business." (copy and paste from Cockroach Labs), then you are competing on managing a server. You are competing with the core business of AWS, and customers will probably find AWS a better choice.
That's not quite so. AWS EC2 provides basically "raw hardware" for you that you have to manage yourself.

You'll need to install the database, hook up monitoring, make sure the infrastructure stays healthy and so on. A fully managed solution typically just provides you with a connection endpoint and offloads all the maintenance to the provider.

This still has some value.

CockroachDB relicensed

Posted Jun 5, 2019 6:45 UTC (Wed) by buchanmilne (guest, #42315) [Link]

>> You could make money from training, consulting, support, bespoke customisations. But if your business model is "Let us manage your database, so you can focus on your business." (copy and paste from Cockroach Labs), then you are competing on managing a server. You are competing with the core business of AWS, and customers will probably find AWS a better choice.

> That's not quite so. AWS EC2 provides basically "raw hardware" for you that you have to manage yourself.
>
> You'll need to install the database, hook up monitoring, make sure the infrastructure stays healthy and so on. A fully managed solution typically just provides you with a connection endpoint and offloads all the maintenance to the provider.
>
> This still has some value.

I don't think timrichardson was referring to EC2 when he said "AWS", probably one of the many other services AWS provides, such as RDS ( https://aws.amazon.com/rds/ ) , AWS Elastic Search (https://aws.amazon.com/elasticsearch-service/), Elasticache ( https://aws.amazon.com/elasticache/ ) etc., which all provide features, making it possible to instantiate and completely manage instances of these services via APIs.

CockroachDB relicensed

Posted Jun 5, 2019 12:08 UTC (Wed) by nilsmeyer (guest, #122604) [Link] (1 responses)

> You could make money from training, consulting, support, bespoke customisations.

Yeah sure, but the revenue here dwarfs in comparison to what AWS makes and offering this as a service on AWS precludes customizations, support and consulting.

> But if your business model is "Let us manage your database, so you can focus on your business." (copy and paste from Cockroach Labs), then you are competing on managing a server. You are competing with the core business of AWS, and customers will probably find AWS a better choice. It didn't work out, so now there is a licence change which charges rent to AWS (or increases the cost of hosting Cockroach DB via AWS, so that the Cockroach Labs offer looks relatively better than now, which is the logic of tariffs).

I think that's a smart move. I don't think users care enough about their freedom to offer a certain piece of software as a service.

> They accepted the rights and obligations of open source. Imagine they found a performance bottleneck, and contributed code to fix it. In return, they received benefits of other contributors. This is a daily scenario for open source, and it shows the rational self interest of open source;

If I contribute to a project that is helpful to me and I receive something back for it that's great for both sides, however I would prefer it if the people who spent a lot of time and money on the project reaping the benefits as opposed to Amazon. The benefits should accrue to the people contributing.

> Ok, that's over the top, but I am far from convinced that AWS is the bad actor here.

I don't think there is a bad actor here, though I'm always surprised at how eager people are to defend Amazon.

CockroachDB relicensed

Posted Jun 6, 2019 6:19 UTC (Thu) by tzafrir (subscriber, #11501) [Link]

> > You could make money from training, consulting, support, bespoke customisations.

> Yeah sure, but the revenue here dwarfs in comparison to what AWS makes and offering this as a service on AWS precludes customizations, support and consulting.

Irrelevant: The real question is: can you make enough (for whatever definition of enough) money out of it. What profits others have from it is not your business.

CockroachDB relicensed

Posted Jun 5, 2019 20:13 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (1 responses)

You're forgetting that cloud giants produce free software too. Quite a lot of it in fact. It's a bit rich to see all those "open source" companies complaining Amazon or Google make a dime with their software, while they see no problem in using kubernetes (for example) without owing a cent to Google.

Amazon or Google or Microsoft will always make more money from hosted open source/free software than small companies. That's not because they are bad open source citizens, that's because they have invested billions in datacenters and the corresponding infrastructure, and that is a *huge* proportion of the value of any hosted service. If you have a problem with that do not make open source hosted service your business model.

All those "open source" companies could close the gap by contributing a little to the basic capabilities of Linux distributions like Debian or Fedora, to diminish the proportion of the value chain provided exclusively by cloud giants, and make it marginal. Do they do that? Of course not. They're in for the money. They laugh at the difficulties of the commons, and then ask you to commiserate when cloud giants behave the same way with them.

CockroachDB relicensed

Posted Jun 5, 2019 20:29 UTC (Wed) by k8to (guest, #15413) [Link]

Classically (earlier this decade), Google and Microsoft produced plenty of open source, but not so much Amazon.

That seems to be changing, a bit.

CockroachDB relicensed

Posted Jun 5, 2019 2:12 UTC (Wed) by dw (subscriber, #12017) [Link] (22 responses)

> They call AWS the "bad guys" when all AWS has done is comply with the licence

In letter perhaps but not in spirit. The _point_ of the license is to encourage community contributions and shared ownership, but that's not how AWS behave. They take the community code, mix it with arbitrary proprietary bits of their own, internalize any and all engineering investment they make into it, stick a price tag on it and compete with the companies who funded the majority of the development on the original code -- often without much more of a differentiator than the ease of billing a near-identical service to a single AWS account.

It's hardly a perfect metric, but I think it's fairly illustrative: RDS for PostgreSQL is around 6 years old by now, yet out of 46k commits to PostgreSQL, only 28 commit messages mention Amazon in any way.

CockroachDB relicensed

Posted Jun 5, 2019 5:32 UTC (Wed) by smurf (subscriber, #17840) [Link] (16 responses)

The point of the Apache license may or may not be to *encourage* community contributions – but it certainly doesn't *require* contributing your enhancements back. That's exactly why companies with lots of resources (read: Amazon) can code arbitrary enhancements and then sell access to that, without the original author benefiting.

If they had used something like the AGPL, Amazon would be forced to open-source its contributions …

CockroachDB relicensed

Posted Jun 5, 2019 5:41 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (5 responses)

> If they had used something like the AGPL, Amazon would be forced to open-source its contributions …
Amazon would simply ignore the product and either build something custom or use a permissively-licensed competitor product.

AGPL has pretty much failed entirely at its goal. There are almost no true large-scale AGPL projects (but plenty of dual-licensed corporate shareware).

CockroachDB relicensed

Posted Jun 5, 2019 8:49 UTC (Wed) by xophos (guest, #75267) [Link] (4 responses)

Nextcloud is doing pretty well...

CockroachDB relicensed

Posted Jun 5, 2019 17:10 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (3 responses)

https://nextcloud.com/pricing/ - it's a corporate shareware.

CockroachDB relicensed

Posted Jun 5, 2019 17:37 UTC (Wed) by rahulsundaram (subscriber, #21946) [Link]

> https://nextcloud.com/pricing/ - it's a corporate shareware.

Nah. That's just pricing for hosted options. You can do your own thing using the software at https://nextcloud.org/

CockroachDB relicensed

Posted Jun 5, 2019 18:06 UTC (Wed) by flussence (guest, #85566) [Link] (1 responses)

Do you release software under licenses that give end users indemnification, support, hosting and SLAs for free?

Does the software stop being libre and become “corporate shareware” if those are paid extras?

CockroachDB relicensed

Posted Jun 5, 2019 18:09 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]

I was mistaken in this particular case. Typically AGPL-ed projects offer a non-AGPL second license, Nextcloud is pretty much the only exception I know.

CockroachDB relicensed

Posted Jun 5, 2019 7:13 UTC (Wed) by pabs (subscriber, #43278) [Link] (8 responses)

The reason that AGPL wouldn't work for these companies is that they *don't* care about getting contributions back (and probably have a disincentive to include contributed features in the AGPL version), they care about pushing people to pay for their the proprietary version, so that they can pay back their VCs.

CockroachDB relicensed

Posted Jun 5, 2019 7:37 UTC (Wed) by mjthayer (guest, #39183) [Link] (7 responses)

> The reason that AGPL wouldn't work for these companies is that they *don't* care about getting contributions back...

So they can sell exceptions to the licence and both sides are happy. Though I fear there might be additional complications, given that no one seems to be actually doing that.

Off-topic, but I wonder why GNU bothered to have separate GPL and AGPL licences. It would have made sense to me from their point of view to say that the user is entitled to receive a copy of the source, and not worry about whether the user is on the other side of a network or the owner of the device the software is running on.

CockroachDB relicensed

Posted Jun 5, 2019 8:50 UTC (Wed) by joib (subscriber, #8541) [Link]

I read somewhere (sorry, forgot where) that during the GPLv3 process many of the free software advocates were pushing for merging GPL and AGPL (or something to that effect). However, corporations that were part of the process, in particular Google, were very strongly opposed to it, and eventually the free software side caved.

CockroachDB relicensed

Posted Jun 5, 2019 10:55 UTC (Wed) by zoobab (guest, #9945) [Link] (4 responses)

"Off-topic, but I wonder why GNU bothered to have separate GPL and AGPL licences."

Google was lobbying against that during the GPLv3 debate, for their own business reasons.

CockroachDB relicensed

Posted Jun 5, 2019 12:42 UTC (Wed) by khim (subscriber, #9252) [Link] (3 responses)

Google is not the only company which have "no AGPL" rule. There's irony, though: in the end Google still went Clang route anyway and is now trying to remove all GPLv3 completely.

CockroachDB relicensed

Posted Jun 11, 2019 9:21 UTC (Tue) by jra (subscriber, #55261) [Link] (2 responses)

That is not true. Google prefers to release under Apache2 but there is no GPLv3 prohibition in Google and GPLv3 code is widely used there (including in new products).

Please don't spread misinformation if you are not in full possession of the facts.

CockroachDB relicensed

Posted Jun 11, 2019 9:49 UTC (Tue) by zoobab (guest, #9945) [Link] (1 responses)

@jra is there a prohibition of AGPLv3 inside Google then? The last article I could find is from 2011 (https://www.theregister.co.uk/2011/03/31/google_on_open_s...), and during some legal panel discussion during GSOC2017, it was pretty clear AGPL was not welcomed for use inside Google.

CockroachDB relicensed

Posted Jun 11, 2019 21:31 UTC (Tue) by jra (subscriber, #55261) [Link]

Yes. aGPLv3 is not allowed. Regular GPLv3 and LGPLv3 is fine.

CockroachDB relicensed

Posted Jun 5, 2019 18:43 UTC (Wed) by andresfreund (subscriber, #69562) [Link]

> Off-topic, but I wonder why GNU bothered to have separate GPL and AGPL licences. It would have made sense to me from their point of view to say that the user is entitled to receive a copy of the source, and not worry about whether the user is on the other side of a network or the owner of the device the software is running on.

From my non-lawyer view - but including talking to a few lawyers about it - the AGPL restrictions are so vaguely formulated that it's hard to know what is actually intended. I mean, clause 13, which does a large fraction of the work for AGPL, is just a mess:

https://www.gnu.org/licenses/agpl-3.0.en.html

13. Remote Network Interaction; Use with the GNU General Public License.

> Notwithstanding any other provision of this License, if you modify the Program, your modified version must prominently offer all users interacting with it remotely through a computer network (if your version supports such interaction) an opportunity to receive the Corresponding Source of your version by providing access to the Corresponding Source from a network server at no charge, through some standard or customary means of facilitating copying of software. This Corresponding Source shall include the Corresponding Source for any work covered by version 3 of the GNU General Public License that is incorporated pursuant to the following paragraph.

> Notwithstanding any other provision of this License, you have permission to link or combine any covered work with a work licensed under version 3 of the GNU General Public License into a single combined work, and to convey the resulting work. The terms of this License will continue to apply to the part which is the covered work, but the work with which it is combined will remain governed by version 3 of the GNU General Public License.

What does "interacting" mean? What does "user" mean? What does "if your version supports such interaction" actually mean? What do the limitations of virality actually mean?

Leaving actual policy goals aside, I'm very happy that the AGPL was *not* merged into normal GPL, just from the perspective of actually understanding what the license means.

CockroachDB relicensed

Posted Jun 5, 2019 18:29 UTC (Wed) by andresfreund (subscriber, #69562) [Link]

> The point of the Apache license may or may not be to *encourage* community contributions – but it certainly doesn't *require* contributing your enhancements back.

That doesn't mean we can't, or shouldn't, criticize them for not contributing. And point out that that approach can be shortsighted etc.

CockroachDB relicensed

Posted Jun 5, 2019 9:06 UTC (Wed) by joib (subscriber, #8541) [Link] (2 responses)

1) It's not only about databases. How much has Amazon contributed to the Linux kernel (answering myself, yes they contributed drivers for their own network NIC, but something generic that benefits everyone?), glibc, gcc, etc. that they rely on every day?

2) It's not only about Amazon. Lots of other companies make $$$ from FOSS software without contributing anything back.

"All software should be free and all that, except for my little piece (that stands on the shoulders of a lot of underlying FOSS software)", isn't a particularly consistent position, and I'm hard pressed to find an enormous amount of sympathy for the latest batch of VC backed startups trying to cash in.

In the end, one shouldn't perhaps except particularly high morals (for whatever definition of morality you may choose) from these corporations, big or small. It's realpolitik all the way, and they'll use whatever arguments and appeals to morality they think makes them the most $$$.

CockroachDB relicensed

Posted Jun 5, 2019 17:34 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link] (1 responses)

> 1) It's not only about databases. How much has Amazon contributed to the Linux kernel (answering myself, yes they contributed drivers for their own network NIC, but something generic that benefits everyone?), glibc, gcc, etc. that they rely on every day?
Amazon employs several kernel developers who are using their own personal emails, so you don't see them in the usual LWN stats. No idea why.

Kernel stats and email addresses

Posted Jun 5, 2019 17:46 UTC (Wed) by corbet (editor, #1) [Link]

We do try pretty hard to correctly attribute contributions regardless of the email addresses used — lots of kernel developers use private email addresses. If you think we're missing somebody, please do let us know.

CockroachDB relicensed

Posted Jun 5, 2019 13:36 UTC (Wed) by martin.langhoff (subscriber, #61417) [Link]

If you pick a permissive license, you are making a statement of your social contract -- it's ok to just take and not contribute back.

GPL makes a different statement -- share back! -- which keeps some (selfish) players away. For sofrware that can be offered as SaaS, the share-back has a big exception.

AGPL makes a stronger share back social contract. Is that what they wanted? AGPL + a commercial license?

Here's the thing -- if you start with a model like AGPL + commercial, you have to build something truly compelling, so that developers gravitate to you and have to choose either AGPL or commercial licensing.

The actual model lately seems to be -- use a permissive license until you're popular, then spring the relicensing trap on your users and contributors. That's f'd -- users of a DB are usually quite tied to it.

If I was a major user of such a software, I'd fork them at the relicensing point.

CockroachDB relicensed

Posted Jun 5, 2019 18:34 UTC (Wed) by andresfreund (subscriber, #69562) [Link]

> It's hardly a perfect metric, but I think it's fairly illustrative: RDS for PostgreSQL is around 6 years old by now, yet out of 46k commits to PostgreSQL, only 28 commit messages mention Amazon in any way.

While I completely agree that Amazon doesn't even remotely do enough, that's not really a meaningful measure. With a few exceptions PG committers (I am one), don't list email addresses of contributors, and employers are just about never listed. We do link to the email discussions by message-id, and that *sometimes* will allow to infer the sender. Even if a contribution is from amazon, they might use a private email, or the most pertinent message linked wasn't from the contributor.

CockroachDB relicensed

Posted Jun 5, 2019 17:43 UTC (Wed) by flussence (guest, #85566) [Link] (2 responses)

Their business assumed the old threat model of some random bad actor with lots of money, like Microsoft, launching a frontal assault against their product via litigation (through fronts like SCO) or buying out (like they did with Skype). Traditional “open source” licenses only care to defend against that.

What we've instead had over the past decade is bad actors with an unfathomably larger amount of money undermining FOSS (and open standards), by running their own fork in parallel until the original dies through attrition. Until licenses catch up and start protecting free software from this angle, the problem's only going to get much worse.

Aside: This is how the Web died, starting with Apple forking KHTML and ending with a single vendor dictating the platform for websites and desktop apps *on every OS*. MS OOXML is looking like a more coherent and implementable spec than “HTML” nowadays - that's no accident.

CockroachDB relicensed

Posted Jun 6, 2019 8:21 UTC (Thu) by nim-nim (subscriber, #34454) [Link] (1 responses)

No, that's not how the web died.

The web died with *users* choosing to rely on a single provider, and expecting someone else to keep the competition alive.

(And, actually, it's not dead yet, though Firefox has huge problems now, which can be traced to years of following slavishly whatever Google wanted browser-side, and expecting Google to construct a web where Firefox was relevant).

CockroachDB relicensed

Posted Jun 6, 2019 22:11 UTC (Thu) by NYKevin (subscriber, #129325) [Link]

Reports of the web's death are greatly exaggerated. It is easier than ever to set up arbitrarily elaborate user interfaces with which users can remotely interact via a web browser, using only standard and open source technologies. The fact that that web browser is usually Chromium-based is a secondary concern or even a non-issue from the web developer's perspective, and in fact, having fewer web browsers makes things easier for web developers in the short term (fewer things to test, no hack-the-HTML-to-make-MSIE-happy, etc.).

Obviously there are long-term problems with having too few choices, and in the short-term, the status quo is not terribly consumer-friendly. We should have more competition in the browser space, now and in the future. But we also want rendering engines to behave similarly to one another for the vast majority of "reasonable" inputs, and that has historically been a much larger problem for the web's health than it is today. (Then: MSIE renders the CSS box model completely wrong. Now: Chrome is shipping semi-niche features faster than other browsers, and it's a little unclear which ones are "standard.")

CockroachDB relicensed

Posted Jun 5, 2019 20:25 UTC (Wed) by k8to (guest, #15413) [Link]

I think typically because the assumption is that no infrastructure will be adopted that isn't open source.

I think this was believed more by the VCs than the company leaders in the last cycle.

CockroachDB relicensed

Posted Jun 6, 2019 19:06 UTC (Thu) by jccleaver (guest, #127418) [Link]

> Why do these firms choose open source licences to begin with, if their monetisation model requires them to restrict use of the code? It looks like they wanted the "sheep's clothing" of open source to build trust, gain contributors and gain market share.

Well, monetization doesn't "require" them to restrict the use of the code when there are norms that downstream users won't intentionally obliterate any way for the originating project to survive. That norm used to be the case, and now it isn't.

> They call AWS the "bad guys" when all AWS has done is comply with the licence. The conflict is not between open source and AWS, it is a conflict between open source and a business model which requires exclusive control of the code.

AWS (and Big Tech in general) has the scale *and the business incentive* to "embrace and extend" any and all OSS project which provides a service. Unlike in prior decades, Amazon, Google et al. don't have to worry about biting the hand that feeds them (smothering the upstream OSS project) because they can thrown essentially unlimited resources at it.

The GPL was written in an era when the operational of the software itself was the key issue, and "open source" meant free access to it and a requirement to give back to the community (in the form of source code) if you were selling it. With the advent of software-as-a-service rather than using-software-to-run-a-website-which-is-the-service, and the vast scale of Big Tech, a new mechanism is needed to ensure that "giving back to the community" remains a relevant concept in the current landscape.

AWS et al have classic oligopoly power due to their scale. Until rectified, any company that wants to pay people to work on OSS projects is at risk of the types of decisions the press release alludes to. Restricting commercial-use-for-external-resale-aaS may be the least-worse option if the alternative is folding if Amazon or Google decide to raise a finger against you.

CockroachDB relicensed

Posted Jun 5, 2019 22:50 UTC (Wed) by fatherlinux (subscriber, #93873) [Link] (2 responses)

Dude, I agree so much. SV money expects explosive growth, while Open Source, when done right, only provides slow and steady growth. Silicon Valley money and Open Source is like oil and vinegar...

CockroachDB relicensed

Posted Jun 6, 2019 7:27 UTC (Thu) by halla (subscriber, #14185) [Link] (1 responses)

Very easy to emulsify?

CockroachDB relicensed

Posted Jun 8, 2019 0:15 UTC (Sat) by ncm (guest, #165) [Link]

Only with surfactants.

CockroachDB relicensed

Posted Jun 4, 2019 22:16 UTC (Tue) by Cyberax (✭ supporter ✭, #52523) [Link] (2 responses)

The question here is, do we actually need new databases?

CockroachDB relicensed

Posted Jun 5, 2019 5:37 UTC (Wed) by edomaur (subscriber, #14520) [Link]

Yes, always

CockroachDB relicensed

Posted Jun 5, 2019 8:02 UTC (Wed) by cthart (guest, #4457) [Link]

Yes, because developers like reinventing the wheel.

CockroachDB relicensed

Posted Jun 5, 2019 5:51 UTC (Wed) by mokki (subscriber, #33200) [Link]

The new license reverts back to apache 3 years after each release. So it is time-shifted open source.

Personally, after reading the blog, I do not feel annoyed at all since I somehow started comparing the new license to AGPL.
I have always liked AGPL from moral point, but practically have never been able to integrate such code to customer projects. The new CockroachDB license is more free to me since it keeps allowing all the same uses as apache license allowed, except for the DBaaS use case, which I have never been interested in building.

CockroachDB relicensed

Posted Jun 5, 2019 13:05 UTC (Wed) by bdelmas (guest, #132196) [Link]

I never used them but what about TiDB[1]? Or FondationDB [2]? They are both licensed under Apache-2.0. Their documentation is not as great as CoachroachDB but they are production ready and under heavy use already.

[1] https://github.com/pingcap/tidb
[2] https://www.foundationdb.org/

How is AWS competitive?

Posted Jun 5, 2019 1:32 UTC (Wed) by scientes (guest, #83068) [Link] (3 responses)

I always felt Amazon was really expensive, I don't get why anyone uses them.

How is AWS competitive?

Posted Jun 5, 2019 5:36 UTC (Wed) by smurf (subscriber, #17840) [Link]

Doing your own hosting is really expensive too, esp. when you're a startup and don't have the $$$$$ required – both for hardware, and for people who know how to design the infrastructure (and then keep it up and running).

How is AWS competitive?

Posted Jun 5, 2019 20:36 UTC (Wed) by k8to (guest, #15413) [Link] (1 responses)

AWS is very expensive for a hosting service.

However, if you want the full resiliency that you can theoretically get out of the entire AWS stack, then building that elsewhere or on your own would be very expensive.

For many customers, the reality falls somewhere in the middle, but they have first-mover shine for a lot of cases.

How is AWS competitive?

Posted Jun 5, 2019 21:00 UTC (Wed) by Cyberax (✭ supporter ✭, #52523) [Link]

> AWS is very expensive for a hosting service.
Lightsail starts at $3.50 per month.

It's really, really hard to find a use-case where self-hosting is going to be cheaper.

CockroachDB relicensed

Posted Jun 5, 2019 8:55 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (5 responses)

CockroachDB is the usual mess produced by non-free-software people, a giant standalone monolithic thing, with a lack of interest in engaging with the community.

It only used open source licensing as a marketing gimmick because their prospective customers like that (the whole community engagement thing, not the fake slap-an-open-source-license-and-behave-the-usual-non-open-source-way).

Hope that will result in purging CockroachDB connectors from actual free software projects, since those connectors require dealing with the CockroachDB codebase.

There are probably a lot more clarifications of this kind in the pipeline, as customers wisen up to the whole open-core/marketing-open-source thing. Entities that avoid doing open source the full way (free software) will lose most of the commercial benefit. Since neither they nor their VCs ever believed in it, reverting to their usual commercial licensing will be the default option.

CockroachDB relicensed

Posted Jun 5, 2019 9:12 UTC (Wed) by joib (subscriber, #8541) [Link] (4 responses)

Well, at least one of the cockroachdb founders and programmers certainly has more free software creds than most of us: https://en.wikipedia.org/wiki/Spencer_Kimball_(computer_programmer)

CockroachDB relicensed

Posted Jun 5, 2019 10:13 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (3 responses)

He has 20-years-old uni free software creds and mostly gave up on it after finishing his degree (nice to have VCs that pay communications agencies to write panegyrics on wikipedia).

And even if he was an outstanding free software citizen in his private life, that does not change how he structured his company.

CockroachDB relicensed

Posted Jun 5, 2019 11:49 UTC (Wed) by nilsmeyer (guest, #122604) [Link] (2 responses)

> And even if he was an outstanding free software citizen in his private life, that does not change how he structured his company.

So he's not allowed to make money?

CockroachDB relicensed

Posted Jun 5, 2019 19:24 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (1 responses)

He can make all the money he wants. And he can do all the free software/open source he wants. Just not achieve the first, by going through the motions of the second, without any serious effort to make the second actually work.

CockroachDB relicensed

Posted Jun 5, 2019 20:39 UTC (Wed) by k8to (guest, #15413) [Link]

Well, obviously he can, but I think you fairly point out that limited credit should be given for doing so.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 5, 2019 13:52 UTC (Wed) by martin.langhoff (subscriber, #61417) [Link] (8 responses)

(surfacing a point I made in an earlier post)

We've seen it with MongoDB. I think there's a pattern that is starting to emerge.

If they cared about freeloaders, they would have chosen a different license -- GPL, AGPL. The founders of Mongo and Cockroach were not naive re licenses at project inception, or if they were, they would have been learned in earlier years.

Using GPL/AGPL still offers a commercial path -- dual license *GPL/commercial. You have to get copyright assignment from contributors; I don't like it personally, but it's not a barrier for this space.

Instead, it seems to me that projects are building up their product and their user/customer base using a permissive license until they reach an inflection point. Factors in reaching the inflection point include AWS offering it under SaaS, scale if market adoption and the interest level of VCs. I might be missing a few.

At the inflection point, it springs the relicensing trap.

I wouldn't want to be a developer today with a codebase tightly tied to CockroachDB. In fact, I'd consider forking them on the spot, at the relicensing point.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 5, 2019 14:38 UTC (Wed) by mathstuf (subscriber, #69389) [Link] (7 responses)

> Using GPL/AGPL still offers a commercial path -- dual license *GPL/commercial. You have to get copyright assignment from contributors; I don't like it personally, but it's not a barrier for this space.

I think if this were set up with a "trap" that if the main project is no longer GPL-available (i.e., taken closed source), it retroactively becomes BSD or some other more permissive license, I'd be much more comfortable contributing to an assignment-oriented project. Take Qt for example: if it ever stops having a FOSS version available, KDAB (or some other entity), gets the latest FOSS release under MIT or BSD license.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 5, 2019 18:56 UTC (Wed) by martin.langhoff (subscriber, #61417) [Link] (6 responses)

Good point. CockroachDB's new 'commercial' license has a clause a bit like Ghostcript. Code is automatically available under a permissive license after 3 years. That earns them bonus points.

I'm still wary of this tide of relicensing. Also -- https://www.scalevp.com/blog/making-sense-of-a-crazy-year...

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 5, 2019 20:26 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (3 responses)

Time shifted open source is pretty useless for software that hasn't reached maturity like ghostscript. It's just another way to claim open source while making your code useless for every one else.

If 3-year old CockroachDB code had any actual value, you can bet that someone would use it without paying CockroachDB anything, and that CockroachDB management would change the licensing again.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 5, 2019 23:56 UTC (Wed) by nilsmeyer (guest, #122604) [Link] (1 responses)

Useless for everyone else? That’s quite the hyperbole.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 6, 2019 6:24 UTC (Thu) by nim-nim (subscriber, #34454) [Link]

It's 100% as useless, as any open core where all the useful bits are segregated in value-added closed modules.

As seen recently with ELK.

Those companies like to give the impression the software is "open source", but if anyone else manages to figure a way to make anything useful out of it (useful enough someone will pay for the service), they will scream betrayal and add another layer of obstruction. Because they just want to claim open sourcing, without the open sourcing to be effective.

They all dream of being the next Oracle, with yachts and an army of compliance auditors bleeding money out of their unfortunate customers. Except Oracle taught those customers never to make their business depend again on a strategic piece of closed software, so they have to give assurances to prospective customers that their software is open (as in, no lock in). Since their actual objective is to set up this lock in, they keep trying to scam their way out of the open obligation.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 6, 2019 9:06 UTC (Thu) by joib (subscriber, #8541) [Link]

Particularly since this is a network service listening on a socket, so being maintained with security updates seems pretty important. Unless I'm mistaken and the company has committed to maintaining those 3 year old versions, at least for a while.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 6, 2019 9:27 UTC (Thu) by joib (subscriber, #8541) [Link] (1 responses)

I do wonder if we're about to see a further bifurcation of the free and open source communities.

With the free software side taking on a more anti-corporate bent, by hackers for hackers, using open source distributed services like mastodon/mediagoblin/etc. rather than mainstream proprietary alternatives. And so forth, all the way to a rejection of modern consumer culture?

And conversely, open source becoming more of a cargo cult marketing gimmick for corporations.

CockroachDB and MongoDB relicensed -- bait and switch?

Posted Jun 7, 2019 0:31 UTC (Fri) by NYKevin (subscriber, #129325) [Link]

OSI has flatly rejected SSPL as an open source license, and to my knowledge CockroachDB has not claimed that their timeshifted license is "open source" at any point before the three years are up. The Commons Clause people are somewhere in between those extremes, as I recall. While the shift you are describing is not an impossible future outcome, I see no evidence of it happening in the broader open source community right now.

CockroachDB relicensed

Posted Jun 8, 2019 8:56 UTC (Sat) by simlo (guest, #10866) [Link]

Any single company open source project can be made non-free in the next version - if copyright is assigned to the company, GPL doesn't help.
But the old version is still open source.

What is missing is a open source community to drive a fork. The problem seems to be that too few users care to put in that effort - and the companies seems to give enough for most people to be satisfied with a half-closed edition.

There is also the case as with GitLab, where essential "enterprise" features are left out in the open source version, but nobody steps up and just make those additions even though they seem trivial: you can obviously not submit those patches to the company, and would have to maintain those out-of-tree.

What for instance FSF should offer, was a platform to help maintain forks of these half open source projects (maybe relicensed to GPL), such users can retain their freedom, when the company change license or won't include obvious features in their original open source version.