LWN's 2021 retrospective

The predictions

The first prediction made in January was that the world would emerge from the depths of the pandemic, and that in-person events would return. Needless to say, things didn't quite work out that way. The pandemic is still very much with us (and possibly about to take another turn for the worse) and, while a few in-person gatherings did take place toward the end of the year, most Linux events are still being held online. The free-software community does still appear to be holding up well, though; it may be true that staying home and interacting with our screens is all we ever wanted to do in the first place.

The prediction that support for CentOS 8 would end was, obviously, obvious; that is still scheduled to happen at the end of this year. Tied to that prediction was a suggestion that, in fact, CentOS 8 Stream might turn out to be good enough for many users, and perhaps even better for some. The lack of "CentOS 8 Stream broke my production system" stories suggests that may have come true, at least to an extent, though it is hard to know for sure.

We also included the prediction that there would be attempts to recreate old-style CentOS 8; given that those attempts were already underway at the time, we cannot claim credit for a lot of foresight. We highlighted Rocky Linux as the highest-profile effort, but lamented its lack of public discussions. Rocky Linux is still out there, and some public mailing lists have been added, but anybody looking for insights in the rocky-devel archives will be disappointed. Meanwhile, AlmaLinux appears to have stolen the spotlight and seems to be doing well, though its communication channels are not particularly friendly to casual browsers. In any case, the prediction that "most or all" of the CentOS 8 recreation efforts were likely to fail does not appear to have been borne out, so far at least.

It is hard to say what became of the prediction that openSUSE would need to better define its relationship with the SUSE mothership; things have been rather silent on that front. The effort to create an independent foundation as the home for openSUSE appears to have stalled, though it would not be surprising to learn that private discussions are ongoing. From what can be seen publicly, nothing has really changed.

Did it become possible to submit kernel patches without touching an email client? In truth, that was possible even before with tools like git send-email. Since then, some further progress has been made; a tool to turn a GitHub pull request into an email series is one relatively prominent example. Coming in the near future will be enhancements to the b4 tool and a new web service that will further remove the necessity for email to submit patches.

So that prediction can be counted as being at least partially successful, but it missed the other half of the equation, which could be expressed as "it will become possible to receive and apply kernel patches without using email". This, of course, refers to the lore+lei work that was recently made public, along with the ongoing work on b4. The kernel community isn't moving away from email anytime soon, but it will become increasingly easy to avoid many of the more annoying aspects of email in kernel development.

Did the commercial side of BPF become more prominent, as predicted? The increasing number of developers working on BPF and the commercial support behind projects like Cilium suggest that the answer is "yes".

The new GNOME 40 interface did make an appearance as predicted, but the expected complaints, for the most part, did not. The changes were not all that disruptive in the end, and it seems certain that most desktop users have, at this point, either made their peace with GNOME or found another solution that is more to their liking.

Included in our predictions was the statement that Python developers would have to think about the future of the language and when it might actually be "done". If long email threads are any evidence, there is clearly some thinking going on, but it still seems to be focused on the shape the new language features should take rather than how many more new features are appropriate at all. At least the prediction that there would be no Python 4 seems to be holding for now.

We predicted that software supply-chain attacks would be a serious threat. That threat is always with us and, on occasion, malicious packages (UAParser, Great Suspender, PHP) were indeed discovered to have been injected into popular repositories. But, as the series of Log4j vulnerabilities makes it clear, our worst enemy in this regard is probably still ourselves. We are quite accomplished at injecting our own vulnerabilities and have no need to outsource that work to outside attackers.

There has been an increase in antitrust enforcement activities, both in the US and Europe, as predicted. As also predicted, things are moving slowly, and there has been little practical effect so far. Also as predicted, the influence of OpenStreetMap continues to grow; the predicted clashes with the hobbyist base appear to have calmed down somewhat instead, though.

What was missed

All told, our 2021 predictions did not go all that badly; arguably that is the result of having not gone too far out on a limb back in January. But there is the associated question of what was missed: what did we fail to predict that we should have perhaps foreseen?

Sometimes the most obvious things can be the hardest to anticipate; consider, for example, the case of minor revision numbers for stable kernel releases. A single byte was set aside for those numbers, since nobody ever thought there would be more than 255 releases of a single stable series. But we live in a different world, with fast-arriving stable updates and kernels that are supported for several years. It should have been possible for us, and for the developers involved, to not only predict that those numbers would overflow, but to make a good guess as to exactly when that would happen. We were all caught by surprise anyway.

The addition of structural pattern matching into Python was a long time in coming with a good chance of happening in 2021. We even mentioned it while talking about Python, but didn't think to predict its acceptance.

It didn't occur to us that Richard Stallman might return to the Free Software Foundation's board of directors, but perhaps it should have. The FSF has always struggled to have an identity separate from its association with Stallman — if, indeed, it wants that at all. So letting him back in may have seemed like the best path forward despite the loud public backlash that resulted.

Perhaps it was not easily predictable that the whole UMN episode, where university researchers intentionally tried to insert buggy patches into the kernel, would happen in 2021. But certainly something like that was going to happen sooner or later; some developers had been warning about the prospect for years. Happily, the kernel's processes worked and, aside from a lot of wasted time, no real harm was done.

Another thing we perhaps should have foreseen but didn't was the intersection of machine-learning technologies and software development. That, too, was only a matter of time; it came to light this year in the form of the GitHub Copilot offering. The controversy over whether Copilot is a violation of free-software licenses has faded over the months, but it seems likely to return as these technologies mature and proliferate.

We have often predicted that the realtime preemption code would be merged into the mainline kernel — and just as often had to make excuses for that prediction when things didn't work out that way. So it is entirely fitting that, in the year when the realtime code really was merged, we didn't even think about predicting it.

Other notes

We lost a number of members of our community this year, including Kent Fredric, Karsten Loesing, Fredrik Lundh, and Jörg Schilling. They will be missed.

This year we produced yet another 50 LWN Weekly Editions; those contained 260 feature articles, of which 227 were written in house. Over 7,200 security alerts went into the weekly editions, along with about 4,300 kernel patches. Our conference coverage reflects the pandemic era, of course, but we still had coverage of eight events in 2021, and assisted with the organization of three of those. It has been yet another busy year — as usual.

At the end of the year we say goodbye to Rebecca Sobol, who is headed off into a well-earned retirement. While she was not present at the very beginning of LWN, she joined us shortly thereafter as the first person who was actually paid to work on LWN, and has been here ever since, even through the times when things looked grim and there was no money for payroll. Over the years she has written articles, attended conferences, herded authors, and interfaced with our group subscription managers; see her goodbye note for details. LWN would not be what it is without her participation, and she, too, will be missed.

There will be some changes at LWN next year as we seek to replace her, but it's not yet clear what form those changes will take; stay tuned. Meanwhile, we wish the best of the year-end holidays for all of our readers; it remains a privilege and an honor to write for, and be supported by, all of you. We will see you all back here next year.