Today I'm asking the #Gentoo arch testers to stop testing stuff using 387 arithmetic. Yep, the one that causes random differences in rounding by using 80-bit registers (vs 64 bits for a regular double), and therefore spams us with useless test errors. Sure, the test suites are broken in the first place by expecting exact results but many upstreams just don't care — and we'd rather focus on real issues. I mean, too often they don't even care about 32-bit arches at all, and bothering them about ancient FPU won't help.

That said, we've already switched the 32-bit multilib builds on amd64 to use `-mfpmath=sse`. The next step would to do the same in new #x86 profiles. While at it, we're also going to need to raise the baseline to SSE2 (e.g. `-march=pentium-m`, `-march=pentium4` or just `-msse2`).

https://www.mail-archive.com/gentoo-dev@lists.gentoo.org/msg98932.html

"""
IN 1999, AFTER ten years of careful work, a researcher at Imperial College in London named Russell Foster proved something that seemed so unlikely that most people refused to believe it. Foster found that our eyes contain a third photoreceptor cell type in addition to the well-known rods and cones. These additional receptors, known as photosensitive retinal ganglion cells, have nothing to do with vision but exist simply to detect brightness — to know when it is daytime and when night. They pass this information on to two tiny bundles of neurons within the brain, roughly the size of a pinhead, embedded in the hypothalamus and known as suprachiasmatic nuclei. These two bundles (one in each hemisphere) control our circadian rhythms. They are the body's alarm clocks. They tell us when to rise and shine and when to call it a day.

[…]

"What's really interesting about these third receptors," Foster told me when we met in his office at Brasenose College, just off the High Street, "is that they function completely independently of sight. As an experiment, we asked a lady who was completely blind — she had lost her rods and cones as a result of a genetic disease — to tell us when she thought the lights in the room were switched on or off. She told us not to be ridiculous because she couldn't see anything, but we asked her to try anyway. It turned out she was right every time. Even though she had no vision — no way of 'seeing' the light — her brain detected it with perfect fidelity at a subliminal level. She was astonished. We all were."
"""

(Bill Bryson, The Body: A Guide for Occupants)

"""
Heat is lost at the surface, so the more surface area you have relative to volume, the harder you must work to stay warm. That means that little creatures have to produce heat more rapidly than large creatures. They must therefore lead completely different lifestyles. An elephant's heart beats just thirty times a minute, a human's sixty, a cow's between fifty and eighty, but a mouse's beats six hundred times a minute — ten times a second. Every day, just to survive, the mouse must eat about 50 percent of its own body weight. We humans, by contrast, need to consume only about 2 percent of our body weight to supply our energy requirements. One area where animals are curiously — almost eerily — uniform is with the number of heartbeats they have in a lifetime. Despite the vast differences in heart rates, nearly all animals have about 800 million heartbeats in them if they live an average life. The exception is humans. We pass 800 million heartbeats after twenty-five years, and just keep on going for another fifty years and 1.6 billion heartbeats or so. It is tempting to attribute this exceptional vigor to some innate superiority on our part, but in fact it is only over the last ten or twelve generations that we have deviated from the standard mammalian pattern thanks to improvements in our life expectancy. For most of our history, 800 million beats per lifetime was about the human average, too.

We could reduce our energy needs considerably if we elected to be cold-blooded. A typical mammal uses about thirty times as much energy in a day as a typical reptile, which means that we must eat every day what a crocodile needs in a month. What we get from this is an ability to leap out of bed in the morning, rather than having to bask on a rock until the sun warms us, and to move about at night or in cold weather, and just to be generally more energetic and responsive than our reptilian counterparts.
"""

(Bill Bryson, The Body: A guide for Occupants)

I'm considering relicensing my projects to #GPL, going forward. Or — at least these projects that involve more lines of code than the GPL copyright notice takes. Why? Perhaps it's just a matter of growing up to realize how bad corporations are. But the more important question is: why did I use permissive licenses in the first place?

Perhaps it was a matter of good nature, a belief in a "permissive" definition of freedom. I wanted my code to help people. It didn't matter to me if somebody else would make money from it, or use it as a part of proprietary software, as long as the original remained free.

Perhaps it was a matter of simplicity — having a short license that I could understand.

Perhaps it was lack of belief in GPL and its enforcement. Things like nVidia repeatedly working around Linux license, grsecurity going proprietary, Oracle's AGPL-based extortion threats or government after government violating OpenSC license. After all, even if some corporation wanted to infringe on my copyright, what could I do?

But I think it's time to change that. Seeing more and more #OpenSource projects go to shit, I think it's time to make a strong statement. To say "I believe in #FreeSoftware, and to hell with corporate exploitation!"

Fun stuff: x11-misc/albert used to be licensed #GPL. Then the maintainer decided to arbitrarily make it proprietary with a custom license ("freeware, i.e. proprietary and source available", with a limited right to redistribute binaries for specific Linux distributions). Except that the project has received some pretty large contributions before that, and the authors of these contributions hold the copyright to them. Since the contributions were made under the GPL, they cannot be incorporated into a proprietary project.

On top of everything, the maintainer has *deleted* the issue discussing the license issues, in particular the GPL violation.

https://github.com/albertlauncher/albert/blob/f9a33001e9e2930291e8d1a8669a6c43d1de2269/LICENSE.md
https://web.archive.org/web/20210225183856/https://github.com/albertlauncher/albert/issues/765
https://bugs.gentoo.org/766129

#Gentoo

I've just merged dist-#kernel eclass changes from Andrew Ammerlaan to #Gentoo. They bring support for "generic UKI" (Unified Kernel Image) kernels. These kernels build a preconfigured "generic" initramfs, then combine it along with the kernel into a single EFI executable.

The change is considered experimental and it will be tested on 6.6.x branch first (either in a future release, or a revbump). It adds a new `generic-uki` flag. If it is disabled (the current default), ebuilds work as usual, install the kernel image and then let installkernel take care of generating initramfs.

When you enable USE=generic-uki, the ebuild will create a generic initramfs, combine it with the kernel into a single UKI executable and install that instead. The postinst phase will afterwards extract the kernel image and initramfs from it, so that (depending on configuration) installkernel can either install the prebuilt UKI executable, the prebuilt initramfs or generate a new one.

We are also going to build our binary kernels in generic-uki configuration. The gentoo-kernel-bin ebuild is going to either install that, or if you disable the flag, extract kernel image from the UKI and install it. On the plus side, this means that we'll eventually be able to provide fully signed kernel images that are suitable for secure boot. On the minus side, this means that the distfiles are going to be larger for everyone.

I'm using the proprietary #mySugr Android app to sync the data from my glucometer (and I'd love to replace it with something open source). One of the app's features is giving points for data input. Enter blood sugar, you get points, enter carbs, points, enter insulin doses, points, tag, points. You get the idea.

Now, insulin doses are split into meal and correction doses. Both grant points separately.

Effectively, this means that if you repeatedly suffer from elevated blood sugars and need to correct them, you get more points than if you're well in range. Makes sense, right?

#diabetes

Today's random set of thoughts, messages and labels:

"Ohmy, the ugly one is riding today! (ED72ac)"
"no peeing"
"cash register malfunction"
"We offer advice on growing potatoes"
"Budzyń - Chodzież. Rolling stock malfunction on train no. 84103/2 (PKP INTERCITY Spółka Akcyjna ), relation Ustka - Bielsko-Biała Główna. Train traffic suspended. We apologize for the disruption."
"So hot, can't sit."
"Door malfunction"
"Train in 9 minutes, and at red [traffic light]"
"Inside the train the AC is cooling but inside the toilet it's heating."