The latest changes implemented in the Systemd repo, related to or prompted by age-verification laws, have made many people unhappy (I suppose links about this aren't necessary). This has led to a surge in Systemd forks during the last days ("surge" because there have always been plenty of forks). Here are some forks that explicitly mention those changes as their reason for forking (rough time ordering taken from the fork page):
-
paramazo/systemd "The systemd System and Service Manager without age verification"
-
ganitam/systemd "Systemd fork just before the Age Verification addition. Hoping more capable developers and maintainers do same.."
-
GSYT-Productions/systemd-fork "The systemd System and Service Manager, without the stupid Age Verification"
-
speedythesnail/unret arded-systemd "The systemd System and Service Manager, without the ret arded age-verification commits"
-
ta13579/systemd "The systemd System and Service Manager WITHOUT THE FUCKING AGE CHECKS"
-
r4shsec/systemd-no-age-verification "This is systemd but without the age verification made via pull request https://github.com/systemd/systemd/pull/40978"
-
Pingasmaster/fightthesystemd "Systemd without the nonsense: no age verification, no lighthouse built-in."
-
Jeffrey-Sardina/system "Liberated systemd -- no surveillance. Ever."
-
HaplessIdiot/systemd-saneagecheck "The systemd System and Service Manager with age verification bypass and polling rate options for said feature"
-
Queer-Coded-LGBTQ/systemd-fuck-california "The systemd System and Service Manager, but without age bs added in."
-
Codiak540/unshitted-systemd "A fork of systemd aiming to strip the Age verification. Sue me california."
Hopefully the energy of this reaction won't be scattered among too many alternatives, although some amount of scattering is always good.
If I saw a speedup that I didn’t understand, then I’d worry that I had accidentally broken something. It’s easy to get speedups by not doing things correctly
Or i’d start looking for backdoors in the old code.
That’s similar to how the backdoor in xz was found. A slightly slower connection caused by obfuscated payloads tipped off a developer to find out what caused the slowdown. His was half a second lag so i’d really be curious what would cause 4.5 seconds.
That’s a lot less likely to be the case; I am aware of just one example of what you describe, and that’s the example you give, whereas I’ve “sped up” my own code many times, by accidentally breaking stuff.
Rather than assume the presence of backdoors, the rational thing is simply to work out why you are seeing a difference in performance, and to determine if you fixed something by accident, or (the more likely scenario) if you broke something by accident