Trust, cancel culture, russia and elbrus2k

[Aliaksey Kandratsenka]

Hi all.

Some of you may have noticed some developments around the aftermath of me amputating basic elbrus2k (aka e2k) support. Prior to recent gperftools release I reverted semi-recently added support for elbrus2k architecture (which I originally reviewed and committed to our git repo; and then reverted). This is (like with most other architectures) just a few lines here and there to just make code compile as our code is a little too unportable in some places. Some even more unportable bits, e.g. heapchecker, wouldn't work anyways.

So what's the matter ? Well, I refuse to review, merge or in general collaborate on anything that is specifically targeting support for SUBJ architecture.

This raises questions that are worth clarifying publicly.

Did anything change in the way I operate as a maintainer? Not really. We're all free software enthusiasts. We're here to have fun and to contribute something back to humanity. As a maintainer I am expected to be a sensible guardian of the code base. That is I should act in the best interest of our users. But the basic principle is that none of us owe anything to anyone. And nobody can be forced or required to work on stuff they don't want. 

I will continue doing the same work as before. I will still do my best for the benefit of our users. But with one exception listed above.

It might seem tempting for me to try to filter who I collaborate with. But this is not how free software works. Even if I tried to exclude contributions from "bad people" whatever that means, it simply wouldn't work.

Similarly, it is not possible to limit who uses our work. Trying to put any barriers or abstain from working on free software would hamper good people more than bad people. I.e. because there are more good people in the world than bad people. So I won't try. In this regard free software is like math or physics. We cannot limit its use for only good uses, whatever that means.

What's the practical effect? Well our project will not build out of the box when targeting this CPU. It will still be trivial for them to fork and apply whatever fixes, and have working binaries. In fact most of the software for this architecture is non-public already. That is, they apparently have a working Linux kernel port, GNU libc and GNU binutils and lots of other stuff (https://disk.yandex.com/d/ZASlXqAM84O1lw), yet none of this is public, it seems. Google-ing around I find that there seem to be NDA of some kind
[https://0x1.tv/Free_software_porting_on_the_Elbrus_architecture_(%D0%90%D0%BD%D0%B4%D1%80%D0%B5%D0%B9_%D0%A1%D0%B0%D0%B2%D1%87%D0%B5%D0%BD%D0%BA%D0%BE,_LVEE-2019)]. There appear to be no way to download e2k binaries or sources for their Linux distro. I was only able to find fairly dated code drops for Linux and binutils-gdb over at github.com/OpenE2K. I.e. they already maintain a fairly large amount of out-of-tree patches. With not much (if any) prospect of landing this upstream. How that doesn't violate GPL, I am not sure. But it is probably sorted out somehow. So adding 5 more lines of "patch" to maintain SUBJ support for gperftools out of the tree would be trivial. Would likely cost them less than the time I spent writing this. Yet it is a matter of principle for me.

So did I betray any trust as a maintainer? Perhaps I have. Indeed this "ban" I have is against the benefit of some users. But my point is, those are not "regular" users, given NDA. And those affected users are exactly the users I want to target with this fairly symbolic limitation.

Lastly, I have every intention of keeping this forum (in the broad sense; that is the github bits and this mailing list) as neutral as possible. I.e. keep the drama away. We're here to develop some free software, not seek divisions. With that in mind, I publicly apologize for calling SUBJ country an evil country in the github ticket. I will keep my opinion off this project.

Thanks for attention and have a nice day.

P.S. Some people over at the "return e2k support" ticket said something about me working for Google. Not sure what was implied exactly, but let me note a couple things. First, this project has no relation with Google for quite some time. Second, I was never paid directly for working on this project. Should be straightforward to verify that I worked on gperftools off working hours. And third, I am a retiree since a few months ago anyways.

P.P.S. Some people started a troll-fest at the "remove x86 support" ticket. I am unsure if this was an attempt of some kind of irony or joke or not. But I take it pretty seriously. Github allows me to block people from our project, so I will.