Monthly Archive:: April 2026
Another major improvement in Redox: a brand new scheduler which improves performance under load considerably. We have replaced the legacy Round Robin scheduler with a Deficit Weighted Round Robin scheduler. Due to this, we finally have a way of assigning different priorities to our Process contexts. When running under light load, you may not notice any difference, but under heavy load the new scheduler outperforms the old one (eg. ~150 FPS gain in the pixelcannon 3D Redox demo, and ~1.5x gain in operations/sec for CPU bound tasks and a similar improvement in responsiveness too (measured through schedrs)). ↫ Akshit Gaur Work is far from over in this area, as they’re now moving on to “replacing the static queue logic with the dynamic lag-calculations of full EEVDF“.
You’d think if there was one corner of the open source world where you wouldn’t find drama it’d be open source office suites, but it turns out we could not have been more wrong. First, there’s The Document Foundation, stewards of LibreOffice, ejecting a ton of LibreOffice contributors. In the ongoing saga of The Document Foundation (TDF), their Membership Committee has decided to eject from membership all Collabora staff and partners. That includes over thirty people who have contributed faithfully to LibreOffice for many years. It is interesting to see a formal meritocracy eject so many, based on unproven legal concerns and guilt by association. This includes seven of the top ten core committers of all time (excluding release engineers) currently working for Collabora Productivity. The move is the culmination of TDF losing a large number of founders from membership over the last few years with: Thorsten Behrens, Jan ‘Kendy’ Holesovsky, Rene Engelhard, Caolan McNamara, Michael Meeks, Cor Nouws and Italo Vignoli no longer members. Of the remaining active founders, three of the last four are paid TDF staff (of whom none are programming on the core code). ↫ Micheal Meeks The end result seems to be that Collabora is effectively forking LibreOffice, which feels like we’re back where we were 15 years ago when LibreOffice forked from OpenOffice. There seems to be a ton of drama and infighting here that I’m not particularly interested in, but it’s sad to see such drama and infighting result in needless complications for developers, end users, and distributors alike. As if this wasn’t enough, there’s also forking drama in OnlyOffice land, the other open source office suite, licensed under the AGPL. This ope source office suite has been forked by Nextcloud and IONOS into Euro-Office, in pursuit of digital sovereignty in the EU. It’s also not an entirely unimportant detail that OnlyOffice is Russian, with most of its developers residing in Russia. Anyway, the OnlyOffice team has not taken this in stride, claiming there’s a violation of the AGPL license going on here, specifically because OnlyOffice adds contradictory attribution terms to the AGPL. It’s a complicated story, but it does seem most experts in this area seem to disagree with OnlyOffice’s interpretation. We’re in for another messy time.
This is the first of a series of articles in which you will learn about what may be one of the silliest, most preventable, and most costly mishaps of the 21st century, where Microsoft all but lost OpenAI, its largest customer, and the trust of the US government. ↫ Axel Rietschin It won’t take long into this series of articles before you start wondering how anyone manages to ship anything at Microsoft. If even half of this is accurate, this company should be placed under some sort of external oversight.
I assume I don’t have to explain the difference between big-endian and little-endian systems to the average OSNews reader, and while most systems are either dual-endian or (most likely) little-endian, it’s still good practice to make sure your code works on both. If you don’t have a big-endian system, though, how do you do that? When programming, it is still important to write code that runs correctly on systems with either byte order (see for example The byte order fallacy). But without access to a big-endian machine, how does one test it? QEMU provides a convenient solution. With its user mode emulation we can easily run a binary on an emulated big-endian system, and we can use GCC to cross-compile to that system. ↫ Hans Wennborg If you want to make sure your code isn’t arbitrarily restricted to little-endian, running a few tests this way is worth it.
I don’t like to cover “current events” very much, but the American government just revealed a truly bewildering policy effectively banning import of new consumer router models. This is ridiculous for many reasons, but if this does indeed come to pass it may be beneficial to learn how to “homebrew” a router. Fortunately, you can make a router out of basically anything resembling a computer. ↫ Noah Bailey I genuinely can’t believe making your own router with Linux or BSD might become a much more widespread thing in the US. I’m not saying it’s a bad thing – it’ll teach some people something new – but it just feels so absurd.
