Soatok Dreamseeker
soatok@furry.engineer
Released some updates for the PKD reference implementations
1
0
0
Soatok Dreamseeker
soatok@furry.engineerI also confirmed that this breaks the duvet traces: https://github.com/fedi-e2ee/pkd-formal/actions/runs/21503512852/job/61954656342
(I'll fix that in the morning.)
1
0
0
Soatok Dreamseeker
soatok@furry.engineerThese changes should actually improve performance.
That might sound weird. "Why would use function and comments make things faster?"
4
0
0
Soatok Dreamseeker
soatok@furry.engineerActually, nevermind: Fixing the duvet stuff seems straightforward.
1
0
0
The goal being to prevent accidental/malicious shadowing of core namespace functions?
The speedup thing is a nice bonus, though :3 I had no idea PHP had execution shortcuts for a bunch of its core functions.
I was originally thinking the speedup ParagonIE was showing was just due to avoiding vtable lookups and was suspicious about how much time was saved lol.
(For those interested, [0] has a good discussion about such "compiler optimized functions" and why you can save so much time. [1] goes a bit deeper.)
[0]: https://tideways.com/profiler/blog/compiler-optimized-php-functions
[1]: https://tideways.com/profiler/blog/we-tried-backslashing-all-the-functions
0
0
0
@soatok
... I also just noticed that in pkd-server-php, every function gets its own `use function` statement, versus pkd-crypto and pkd-client-php, which just use a list of function names.
I'd be happy to spin up a PR to change these over, if it's something that matters to you at all. It tickles my OCD, but ultimately is a nonissue ^^;
1
0
0
/Cinny 
Soatok Dreamseeker
soatok@furry.engineer@indig0 Yeah, I need to clean it up and be more consistent. I was trying to write a php-cs-fixer custom rule but got distracted.
0
0
0
