I like Andrew, but I think that this way of telling the story does not match the reality of the BSD license. With the BSD license, there are many potential forks of a project (with potentially different licenses) that compete at the same time in the "market" of the user base. Who puts the best developments, leads the race.
When I stepped back from Redis, I formally left it in the hands of the Core Team the was constituted by Redis Inc. (the one that included three Redis Labs programmers, one from AWS, one from Alibaba), but this was a formal act from my side: with the BSD license any third party can fork, change name, continue the project with a different trajectory and with different goals.
BSD code is a common good, and everybody can impress new directions, if willing to do so.
Hi antirez. I like you too. I essentially agree with you, but I think you are underestimating the importance of network effects of users herding to one particular branch of the tree.
My blog post is an appeal for users to choose one branch over another.
This ignores the fact that the brand matters, though. If I fork postgres, nobody is going to just start using my fork over the official software, because the two aren't nearly the same. One has people behind it, the other doesn't.
Yes, it may end up that Redict has the backing of the community, and Redis falls behind, but we shouldn't act like all the mindshare isn't currently with Redis (the brand).
Branding is important but it's not Coca-Cola and we are not normal customers of something. I believe that in the end who wins is who provides technical excellence and a license that works for the use case. The reality is that most of the times nobody wants to put substantial work into OSS projects for free (I understand the reasons), and I don't agree with this post when Andy says that projects can be set in maintenance mode just doing minimal work to take them frozen, even when they are mature projects.
Moreover the current situation with Redis is not closed-but-maintained VS forks-possibily-without-development-force, because the SSPL is a quite open license. To put our destiny in the hand of what OSI approves or not is crazy IMHO. Actually for OSS to be viable in the future, we likely need many other licenses based on copyleft, that allows new projects developers to say, OK, I can use an open license because I have some protection from megacorps. Don't get fooled by a minority of strong voices, if you check Hacker News comments about Redis licese change, you will see that at this point many, many folks get that.
Stallman was now almost erased from collective memory but with GPL what happened was exactly what was happening now with SSPL. Initially copyleft was used, and later stronger copyleft was used in successive versions / modifications of the license, like AGPL. So this problem is not new and people a the core of the Free Software movement addressed them in a very similar way decades ago. But now suddenly if a small company goes along the same lines, that is a violation of everything, while big companies like Amazon posing a major threat to future of OSS is fine.
Disclosures: I work for Amazon, but my positions as a long time Free Software advocate and copyleft supporter are my own. I do not speak for Amazon here. I served on the board of directors for Software Freedom Conservancy at its inception, but now I am only a financial supporter in an individual capacity. I do not speak for SFC either.
The SSPL isn’t the same as the GPL. The GPL is a tool that was invented to advance the Free Software movement agenda: protecting the freedoms of users of software. The SSPL does not do that. It protects the interests of companies that want to maintain exclusivity in offering software services.
My parallel is between AGPL and SSPL. The SSPL protects from companies like Amazon: it can protect other companies, or can protect hackers that don't want to see their work exploited (and make zero profits from their stuff). It's a tool. I don't know if I would use it, but it makes totally sense IMHO.
It's simple to see things from the POV of other smaller but yet VCs-backed companies only, when evaluating the damage of Amazon. Things are actually more complex. For instance OpenRedis was one of the very first Redis SAAS services ran by the same folks that contributed to Redis in the first days, that provided the logo and so forth. Guess who is also impacted by the Amazon monopoly?
The needs of those who want to capture the majority of the value they create when they author digital goods like software haven’t been the concern of either the Free Software movement nor the Open Source movement. These movements are rooted in the nature of digital goods: they cannot be “exploited” (like strip mining for extracting materials from a mountaintop does) because they are naturally abundant. “Create more value than you capture” was the mantra.
The “free rider problem” has been used to justify licenses that protect the interests of software authors since before we called it “open source”, long before Amazon was made a scapegoat.
“The free rider problem is when someone is allowed to package free software in non-free or less-free bundles, and that's precisely the area of the GPL that I thought I needed to do something about in making the Aladdin license”—Peter Deutsch, October 1998 [1]
The point I am trying to make is that legal tools like the Aladdin Free Public License and SSPL do not protect FOSS, or advance the social movements that have produced a bountiful commons of digital public goods. Those that craft those tools and apply them to their works have every right to do so, but they do so because they choose to do so, not because a company like Amazon “forced them” to.
The SSPL is useful exactly for the reasons you outlined, mainly. To preserve the ability to make money of the original authors is just a side effect. The SSPL point from the point of view of the Free Software Movement is (whatever was the reason for the creation of such license): people will start to write less and less free software (it is happening already) if they see their software exploited by megacorps. As less free system software is produced, the society captures less and less value. It's exactly the same idea as AGPL. Also if the original authors have a viable income, they no longer need to do things as a side project, and can write more and better free software, so also to avoid Amazon stealing the potential money output is a direct way to maximize the free software main goal.
There is more Free Software produced today than ever before, even if some firms are deciding they no longer want to produce it. There are far more who are continuing to add to the commons of public software unnoticed; those who publicly quit get far more attention.
I’m all about increasing the private provision of public goods. I don’t think that licenses that give firms exclusive rights to monetize the goods they produce advances that goal. That’s just the status quo of firms being landlords of their intellectual property, extracting rents from others who want to enjoy them. That’s just proprietary software, which is clearly a useful thing for society to have.
The thing that leaves a bitter taste in the mouth is when you apply your resources to produce common goods, and then one party declares that you are somehow a “bad actor” while changing the terms to enclose the goods and exclude you from enjoying that common good.
Luckily with FOSS, those who are excluded can move their efforts elsewhere. But the future goods will not be the same. And I think that’s a shame.
SSPL protects vendors from competition of likes of Amazon, cool, but is this what Open Source is about ? I belive first and foremost Open Source (term) is first and foremost about value for Users of Open Source Software.
As a vendor/developer you surely can chose not to provide software users some freedoms, but when lets not call it Open Source
I agree with basically everything you say, and I agree that developers should be able to choose the SSPL, because usually the choice is between SSPL-and-maintained and open-but-languishing.
I don't think you can say "yes but you can just fork it", because that ignores the reality that, sure, maybe eventually that fork will win, but right now the main fork (Redis) has a huge head start.
Con't comment on the fact that what was promised was not maintained, since I'm away for four years at this point. But I believe that in business sometimes, as the situation changes, or correction for monopoly positions are not likely to be taken by governments in the near future, changes are more or less forced. This is just speculation on my side as I was very little involved in the business side already when I was part of the company. Now zero involvement for four years. But the fact here is that many OSS projects needed to go towards this path one after the other. So I believe it is more useful to understand why they are willing to take this risk, what are the conditions that determine the bizarre fact that the main player in the development takes a marginal part of the cut.
I agree, that's why I don't ever want to say that something "will always be X". Otherwise people might be upset when I have to break the promise.
I'm not saying this to you as the creator of Redis, by the way (you already said you've been away too long for that), just as one random HN commenter to another.
Yep, these are problematic statements, since what they actually mean is "as long as I'm in charge" from the POV of project managers, CEOs, ... depending on the context. And even when people don't change conditions change. Indeed it is much better to say something like: we will take the BSD license as long as possible. And actually it must be observed that Redis was one of the latest to change the license.
Yeah, exactly. And people take them at face value ("you promised it'd always be so"), without realizing that, over long enough time frames, all promises will be broken.
I guess on one hand, be careful what you promise, on the other, be careful what you believe.
The "Renamed To" was bothering me (like some other posters mentioned), and then it hit me why:
Forks of large projects happen when something bad is going with the source (why fork something huge when things are going ok?), and marks the source as poisoned.
Renamed means... It just changed name, everything is already, nothing to see here...
Sure there is the mindshare on the name but people in IT for long knows what means when a fork of this size means.
Shouldn't be an issue for cloud SaaS. For self-hosted they'll have to provide an object file of the software minus the LGPL portion; that preserves the IP, it's more a bit of annoyance w.r.t. an extra step for the build process and distribution.
In the usual way, yes you are correct. I think the article is trying to present an alternative viewpoint, where the real Redis product is not the one that is still named "Redis" but the one that continues being opensource.
So it is a fork, but we might as well accept that what is currently named Redis will slowly (or quickly?) morph into something unrecognizable, what with the focus on generative AI and commercialization...
Uh, this article is so strangely written. The real thing that happened is https://codeberg.org/redict/redict — a fork of Redis by Drew DeVault under LGPL v3.
I don’t think I’m happy with the licence choice here (it should stay BSD-3-Clause). At least the primary repository isn’t on sourcehut, people might actually contribute to it.
> I don’t think I’m happy with the licence choice here (it should stay BSD-3-Clause). At least the primary repository isn’t on sourcehut, people might actually contribute to it.
The old development community is regrouping under https://github.com/placeholderkv/placeholderkv. As of right now we don't intend to change the license, we just want a more balanced governance that prevents a corporate entity to make this type of choice. We are currently planning a five person governance, with everyone coming from a different company. (We would love some independent maintainers, but there weren't any at the time
the article isn't strange at all, it's pointing out that there are two projects based on the last-good-release-of-redis source code:
1. redict, maintained by drew de vault, which takes the stance that redis is finished software, works well, and just needs its existing feature set maintained and bug fixed
2. redis, maintained by redis labs, which seems to be experimenting with generative ai for the sake of adding features, and which practically needs to stay on the feature churn treadmill so that they can keep trying to extract a profit from it
the author's (quite reasonable) position is that regardless of who gets to use the name "redis", the product that people know and love is going to live on as drew's branch of the fork, not redis labs'.
The article ignores that the two previous core team members that didn’t work for Redis Ltd, along with other active contributors, are gathering around a different project. They just don’t have a name yet.
But why is that a "quite reasonable" position? To me, it's the opposite: extremely unreasonable, bordering on ridiculous and over-dramatic. Consider:
* This position conflates a license to use a product with the product itself. That's not how intellectual property works!
* The blog post author is unilaterally speaking on behalf of a community, but it isn't clear that the author has ever contributed to Redis (has he?) or is a major part of the Redis community (is he?). I understand that he is a very accomplished coder and have heard many positive things about Zig, but where's the relation to Redis?
* Ditto for Drew (maintainer of the fork in question), has he ever contributed to Redis or is he a major part of the Redis community?
fwiw some people said this same thing about MySQL vs MariaDB, claiming (absolutely incorrectly) that Oracle's MySQL is the "fork". In that case, it wasn't a license change (MySQL is still GPL), but rather people that just didn't like the change in ownership. So it seems people will just invent mental gymnastics to justify this "X is really just a fork of X" when they don't like a decision by the intellectual property owner.
In the MySQL/MariaDB case, the people saying this "MySQL is actually the fork now" were largely not terribly visible members of the community. And the Linux distributions who mapped the "mysql" package onto "mariadb" created a giant mess, since MySQL has remained more popular than MariaDB, and the two are absolutely not compatible with each other.
Anyway, if you don't like Redis Ltd or what they've done here, that is fine and understandable! But it doesn't mean you can unilaterally declare that their product is a fork of itself, and the "real" project lives on elsewhere, and expect most people to go along with this as if it is a reasonable position.
you're right, it's not really reasonable to say redict is the only valid fork, especially since other people have pointed out that some of the original redis devs have their own fork. i had the impression there was more community behind drew's fork then there actually is.
The license choice is interesting here - this means if there are many committers and there is no requirements like CLA it becomes impossible to change the license and have proprietary forks without all contributors agreements.
I agree though I would rather see them choose to stick to BSD and work on creating community of vendors (rather than Redis Labs monopoly) perhaps some purely Open Source, others proprietary.
To me, the article presents an alternate narrative that (in my opinion) rightly calls out relicensing of F/OSS as the actual fork since it’s being done in a (usually) hostile manner against the wishes of the community that lifted it to be relevant in the first place.
I think the crack at Drew is (maybe) fair given the explanation of redemption, but also unnecessary. Drew’s hardline software freedom advocacy is the relevant information to the story. His promise of redict’s safe harbor is backed up by his profound stewardship of other projects.
Redis is the original project and it continues, even if the licence change is a bad move. Redict is a brand new fork, currently at the “rename everything” stage. It may or may not succeed at being the “new Redis”. The people behind “old Redis” probably consider this title misleading and possibly defaming their project. Many people may decide to create other forks (I expect OpenRedis from Amazon any second now).
For many people, Redis is an open-source in-memory data store. Redict is closer both in spirit and in legal usability to the Redis people have been using for many years compared to whatever got the Redis trademark now.
If I do an `apt-get upgrade`, I don't expect to schedule calls with lawyers. I imagine most repos will switch to either this one or some other "renamed" projects, with the package name "redis" pointing to one of those.
I knew Redis Labs was a sleazy outfit back when they falsely claimed on billboards to be the “Home of Redis”. That they hired Antirez a few years ago and eventually bought Redis did not change this. We need to converge on one fork to supplant theirs, and redict seems as good as any. Unfortunately redict.com is squatted by DNS speculators.
Cynically a bunch of, idk, clout chasers maybe or opportunists?, are going to try to fork to advance them selves and take advantage of some uncertainty? What does this guy know about redis?
Agree with the other comment, this was written weirdly. Can't place my finger on why though.
Drew seems like a great guy, but he wasn't the one maintaining Redis when the relicense happened. There are non-Redis contributors who are trying to figure out what to do nest, and people keep peppering me (one of those contributors) about this fork. I like the meta-point Andrew makes that Redis is the one that's forking, but the framing it around some other arbitrary fork doesn't sit well with me.
When developers do this is it really for the good of the community or is it a career advancing move?
Because the community has a bunch of redis protocol compatible choices and I don't understand why pushing redis in particular over others is the right choice.
> When developers do this is it really for the good of the community or is it a career advancing move?
Drew had the opportunity to make money with this fork by hosting it on SourceHut, and potentially creating subscribers out of redict contributors. Instead, he chose to host on Codeberg, a German non-profit, that is more GitHub like. I think, this alone, shows that (in this case) it’s about community and preservation over everything else.
I’d still expect the size of redict’s community to be big enough to convert some folks to paid. But maybe I am overestimating the size of that community since the Redis company took over?
I didn’t think you were commenting on Drew, fwiw. This keeps happening, and the people stepping up sometimes have different motivations. I believe Drew’s are well intentioned, and am pointing it out as “I think it’s possible to go well.”
When I stepped back from Redis, I formally left it in the hands of the Core Team the was constituted by Redis Inc. (the one that included three Redis Labs programmers, one from AWS, one from Alibaba), but this was a formal act from my side: with the BSD license any third party can fork, change name, continue the project with a different trajectory and with different goals.
BSD code is a common good, and everybody can impress new directions, if willing to do so.