Linux creator Linus Torvalds takes leave, apologizes for his behavior

[ So this email got a lot longer than I initially thought it would get, but let’s start out with the “regular Sunday release” part]

Another week, another rc.

Nothing particularly odd from a technical point of view in last week's kernel updates – rc4 is fairly average in size for this stage of the release cycle, and all other statistics also seem normal.

We have about two-thirds of driver patches (GPU and networking seem to be essential, but the changes are minimal in various driver subsystems), the rest being the usual mix: networking main, tool updates, Documentation, some file systems, vm fixes and minor kernel.

So everything is pretty small and normal for this step. As usual, I add the shortlog at the bottom for those who want to have an overview of the details without having to go into the tree.

The only change that stands out and that deserves to be mentioned is the addition of the code of conduct …

[ And here comes the other, much longer, part… ]

Which brings me to the * NOT * part of last week: the discussions (in public mostly on the top kernel mailing lists, then a lot in various private communications) about maintenance and the kernel community. Part of this discussion was caused by the fact that my planning for the maintainer's summit was started and that these questions are supposed to be discussed.

And do not get me wrong. It's not as if the discussion itself was new this week – we've been discussing sustainment and the community for years. We had a lot of discussions in private and on the mailing lists. We have regular talks at conferences – again, the type "public speaking" and the type "private hallway".

No, what was new last week is really my reaction to that, and I may be introspective (you are the judge).

There were two parts to that.

One was simply my own reaction to having messed up my top maintenance planning: yes, I was a little embarrassed to have messed up my schedule, but honestly, I especially hoped not to have not having to go to the top of the core I have been going every year for about the last two decades.

Yes, we reprogrammed it, and no, my "maybe you can just do it without me there" was rejected. But this whole situation then began another type of discussion. And somehow to that, the second part was that I realized that I had completely misread some of the people involved.

It's where the moment "look in the mirror" comes in.

So here, finally, on the one hand, realizing that it was not really funny or a good sign that I was just hoping to completely skip the annual kernel summit, and on the other hand realize that I had really ignored feelings deep in the community.

It's one thing when you can ignore these problems. Usually, it's just something I did not want to deal with.

That's my reality. I am not an emotionally empathic person and this will probably not surprise anyone. The least of me The fact that I misunderstood people and that I do not realize (for years) how much I've judged a situation and contributed to a non-professional environment is not a good thing.

This week, people in our community have been confronting me about my life for not understanding emotions. My flippant attacks in e-mails have been both unprofessional and useless. Especially when I made it personal. In my quest for a better patch, it made sense to me. I know now that it was not ok and I am really sorry.

The above is fundamentally a lengthy way to arrive at a somewhat painful personal warning that I must change some of my behaviors, and I would like to apologize to people that my personal behavior has hurt the development of core. .

I will leave and get help to understand people's emotions and react appropriately.

In other words: When I was invited to conferences, I sometimes said that kernel development problems did not usually relate to technical problems, but rather inflection points where the flow development and behavior changed.

These issues have been related to patch flow management and have often been associated with major tool changes – from the publication of "patches and tarballs" (and to the very painful discussions about how "Linus does not evolve "). + years ago) to use BitKeeper, then having to write git to get past the point of no longer working for us.

We have not had that kind of pain in a decade or so. But this week, I felt this kind of pain.

To link all this to the current version 4.19-rc4 (no, in fact, this _is_ is linked!) I actually think that 4.19 seems pretty good, things have come to the "quiet" period of the release cycle, and j & I spoke to Greg to ask him if he would be ready to finish 4.19 for me so that I could pause, and at least try to repair my own behavior.

This is not some kind of "I'm exhausted, I just have to leave". I do not feel like continuing to maintain Linux. On the contrary. I * really want to * continue to do this project I have been working on for almost three decades.

It's more like the time I came out of kernel development for a while, because I had to write a little tool called "git". I need to pause to get help on how to behave differently and to solve some problems with my tools and my workflow.

And yes, some could be "just" tools. Maybe I will be able to set up an e-mail filter, so when I send an e-mail with evil words, they will not come out. Because I am a big proponent of tools, some automatisms can at least improve certain problems.

I know that when I really look "in the mirror" it will be clear that this is not the only change that needs to happen, but hey … You can send me suggestions by email.

I look forward to seeing you at the Maintainer Summit.

Linus