Hi all, A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set. This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community. Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central. Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here. Cheers, Bert [1]: https://lists.archlinux.org/archives/list/arch-dev-public@lists.archlinux.or... [2]: https://antirez.com/news/151 [3]: https://redis.io/blog/agplv3/
On 5/2/25 5:04 PM, Bert Peters wrote:
This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
Hi. As you say, this sounds like a pragmatic way forward! //Torxed
On Fri, May 02, 2025 at 05:04:51PM +0200, Bert Peters wrote:
Hi all,
A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set.
This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community.
Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
This has already been discussed a little bit on IRC but I'll reiterate my points here, along with a couple more points. My opinion is that we should still drop redis. It should be clear to us at this point that Redis Inc is not a reliable upstream, nor company, to deal with. And thus simply not worth our volunteer time and efforts to support, promote and maintain. They are still breaking the terms of the BSD license, at the time of writing, and they have no redeemed themself in anyway. The upstream is now licensed under three different licenses, along with a CLA, that allows them to pull off any further shenanigans as they see fit. They are simply not worth our time and effort, and we are better off promoting projects that actually do serve our community. -- Morten Linderud PGP: 9C02FF419FECBE16
On 5/2/25 11:04 AM, Bert Peters wrote:
Hi all,
A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set.
This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community.
Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
Cheers,
Bert
[1]: https://lists.archlinux.org/archives/list/arch-dev-public@lists.archlinux.or... [2]: https://antirez.com/news/151 [3]: https://redis.io/blog/agplv3/
At first I thought that we should continue with the removal so as not to encourage what the upstream has done, but now after giving it some more thought, I don't like the precedent that that would set. If someone wants to maintain Redis in [extra], they should feel free to do so, but we should still patch all packages to use one of the properly open source alternatives in all possible cases. In an ideal situation, Redis remains in [extra] with no revdeps so it can be immediately removed if any further upstream footguns occur. Campbell Jones
On 5/2/25 5:04 PM, Bert Peters wrote:
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
I think as long as 1) there's nothing preventing us from packaging it and 2) there's a maintainer willing to take care of it, it should be fine to have in the repos. The suggestion to not have another package depend on it, or having some other package have a `provides=(redis)` seems very reasonable though. cheers, kpcyrd
On Fri, May 02, 2025 at 05:04:51PM +0200, Bert Peters wrote:
Hi all,
A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set.
This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community.
Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
Cheers,
Bert
[1]: https://lists.archlinux.org/archives/list/arch-dev-public@lists.archlinux.or... [2]: https://antirez.com/news/151 [3]: https://redis.io/blog/agplv3/
Hi Bert, First of all, thank you for bringing the IRC/Matrix discussion to the ML! Whatever led redis to reconsider, the way they treated us is unacceptable to me as a volunteer package maintainer. Anthraxx, then Andrew and I got contacted by a product manager from Redis. There were no apologies for the licensing issues, nor any for future coorperation - only some information about the latest Redis features and a statement that removing Redis from the official repos would not be in the interest of "the community" (whatever community is meant in this context). In contract, we've had very positive interactions with the Valkey upstream. There is a clear interest in making Valkey packaging easier for distributions. For example, we'll likely be able to remove the jemalloc patch that makes valkey/redis link against the system jemalloc in the future. So how do we proceed? I agree with Morten, that Redis upstream has proven itself abolutely unreliable. I propose we exclude redis from the official repos for one year, after which we can reevaluate its status. Contining to aintain Redis will only result in more unnecessary drama. We've already announced Redis's replacement and we should stand by that decision rather than follow upstream’s erratic direction. Best regards, Frederik
On May 3, 2025 10:09:19 AM EDT, Frederik Schwan <freswa@archlinux.org> wrote:
On Fri, May 02, 2025 at 05:04:51PM +0200, Bert Peters wrote:
Hi all,
A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set.
This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community.
Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
Cheers,
Bert
[1]: https://lists.archlinux.org/archives/list/arch-dev-public@lists.archlinux.or... [2]: https://antirez.com/news/151 [3]: https://redis.io/blog/agplv3/
Hi Bert, First of all, thank you for bringing the IRC/Matrix discussion to the ML!
Whatever led redis to reconsider, the way they treated us is unacceptable to me as a volunteer package maintainer. Anthraxx, then Andrew and I got contacted by a product manager from Redis. There were no apologies for the licensing issues, nor any for future coorperation - only some information about the latest Redis features and a statement that removing Redis from the official repos would not be in the interest of "the community" (whatever community is meant in this context).
In contract, we've had very positive interactions with the Valkey upstream. There is a clear interest in making Valkey packaging easier for distributions. For example, we'll likely be able to remove the jemalloc patch that makes valkey/redis link against the system jemalloc in the future.
So how do we proceed? I agree with Morten, that Redis upstream has proven itself abolutely unreliable. I propose we exclude redis from the official repos for one year, after which we can reevaluate its status. Contining to aintain Redis will only result in more unnecessary drama. We've already announced Redis's replacement and we should stand by that decision rather than follow upstream’s erratic direction.
Best regards, Frederik
I wasn't aware of the Redis mails to you/Morten/Levente. With that context, I agree with your suggestion of a one-year ban from [extra]. We've already made the announcement and preparations for the removal and Redis should reap what they've sown, so to speak. Campbell
On May 3, 2025 1:24:31 PM EDT, Campbell Jones <serebit@archlinux.org> wrote:
On May 3, 2025 10:09:19 AM EDT, Frederik Schwan <freswa@archlinux.org> wrote:
On Fri, May 02, 2025 at 05:04:51PM +0200, Bert Peters wrote:
Hi all,
A while ago we as a distro decided to stop supporting Redis, due to their licence change [1], and move to Valkey. Through a combined effort, we removed all direct dependencies on redis, replacing them with vault and patching until that worked. The announcement was posted, and a deadline was set.
This may or may not have caused Redis to reconsider their license change, and have announced another relicencing, this time to the AGPL [2] [3]. With that change, I personally believe there is no longer a reason to remove redis from [extra], and keep it in the repos as-is. Redis is almost but not quite compatible with Valkey, so dropping it without good cause would be a disservice to our community.
Now, I don't want to make light of the harm that Redis inc initially wrought on the open source community with their license change, or waste the work that was done to make everything work with Valkey. Pushback like this is what caused the license change. As such, I propose we continue to use Valkey as the implementation for all purposes that don't strictly require Redis, and maintain Redis simply as a package for our commynity's convenience. That way, should the licensing change again in the future, we do not have a similar amount of work ahead of us. This seems to me the Arch way: pragmatic and user central.
Now I know that this is not a universally shared opinion, so please consider this email an invitation for discussion on what we should be doing here.
Cheers,
Bert
[1]: https://lists.archlinux.org/archives/list/arch-dev-public@lists.archlinux.or... [2]: https://antirez.com/news/151 [3]: https://redis.io/blog/agplv3/
Hi Bert, First of all, thank you for bringing the IRC/Matrix discussion to the ML!
Whatever led redis to reconsider, the way they treated us is unacceptable to me as a volunteer package maintainer. Anthraxx, then Andrew and I got contacted by a product manager from Redis. There were no apologies for the licensing issues, nor any for future coorperation - only some information about the latest Redis features and a statement that removing Redis from the official repos would not be in the interest of "the community" (whatever community is meant in this context).
In contract, we've had very positive interactions with the Valkey upstream. There is a clear interest in making Valkey packaging easier for distributions. For example, we'll likely be able to remove the jemalloc patch that makes valkey/redis link against the system jemalloc in the future.
So how do we proceed? I agree with Morten, that Redis upstream has proven itself abolutely unreliable. I propose we exclude redis from the official repos for one year, after which we can reevaluate its status. Contining to aintain Redis will only result in more unnecessary drama. We've already announced Redis's replacement and we should stand by that decision rather than follow upstream’s erratic direction.
Best regards, Frederik
I wasn't aware of the Redis mails to you/Morten/Levente. With that context, I agree with your suggestion of a one-year ban from [extra]. We've already made the announcement and preparations for the removal and Redis should reap what they've sown, so to speak.
Campbell
Correction, you/Andrew/Levente. I was thinking about Morten's response while writing.
Hi, On 03/05/2025 15:09, Frederik Schwan wrote:
So how do we proceed? I agree with Morten, that Redis upstream has proven itself abolutely unreliable. I propose we exclude redis from the official repos for one year, after which we can reevaluate its status. Contining to aintain Redis will only result in more unnecessary drama. We've already announced Redis's replacement and we should stand by that decision rather than follow upstream’s erratic direction.
I agree with this. Ban and revisit in the future if upstream is better and there's appetite from Package Maintainers to maintain. Cheers, -- Leonidas Spyropoulos Developer & DevOps PGP: 59E43E106B247368 244740D17C7FD0EC
Hi all, The list has gone quiet, so I think it's appropriate to gauge sentiment and I'm going to do so in the only way I have: tallying comments. A few different options have been proposed * Keep it in the repos, avoid depending on it going forward. This is supported by Torxed, kpcyrd, and myself. * Ban it for a year, revisit the discussion later to see how upstream has kept their promises. This is supported by Freswa, serebit, and artafinde. * Drop redis, with no plans for re-adding it. It can live in the AUR. This is supported by Foxboron. Overall, I see a generally negative sentiment towards keeping it in the repos. As such, I will not push this issue any further. While this is in no sense an official vote (and we didn't reach any level of participation to make it so), I think it's best to continue with the drop as originally planned. Thank you all for your thoughts and entertaining my proposal, Bert.
On Mon May 12, 2025 at 11:26 AM PDT, Bert Peters wrote:
Hi all,
The list has gone quiet, so I think it's appropriate to gauge sentiment and I'm going to do so in the only way I have: tallying comments. A few different options have been proposed
* Keep it in the repos, avoid depending on it going forward. This is supported by Torxed, kpcyrd, and myself. * Ban it for a year, revisit the discussion later to see how upstream has kept their promises. This is supported by Freswa, serebit, and artafinde. * Drop redis, with no plans for re-adding it. It can live in the AUR. This is supported by Foxboron.
Overall, I see a generally negative sentiment towards keeping it in the repos. As such, I will not push this issue any further. While this is in no sense an official vote (and we didn't reach any level of participation to make it so), I think it's best to continue with the drop as originally planned.
Thank you all for your thoughts and entertaining my proposal,
Bert.
Thanks for a calm and rational approach to this. :)
participants (8)
-
Anton Hvornum
-
Bert Peters
-
Brett Cornwall
-
Campbell Jones
-
Frederik Schwan
-
kpcyrd
-
Leonidas Spyropoulos
-
Morten Linderud