@s0yuka And in parallel we patiently wait for xdebug to do an awesome job in supporting PHP 8.5, in order to thereafter also give the api platform templates some live.
@VotrubaT Not preferring Singleton is a great way to obfuscate facades even further, as you then have even less of a clue which class instance you are dealing with. Pro: allows implementations to make bc breaks, as noone know which implementation is used. Could be an anti corruption layer.
@s0yuka@ApiPlatform Where can we help? Am preparing what's needed to start working on a prototype implementation for this at the end of this month myself.
@markusstaab@mangledObject@wdesportes Nice! Is there a difference in feature parity between PHPs reflection API and https://t.co/FXI0KgW09I that was used before? Guess PHP's is sufficient for downgrading 8.2 and lower to 7.4? What would it take to cover 8.3 as well?
@markusstaab@OndrejMirtes@Ruud_@VotrubaT Haha my attempt on a bad joke, with 0.9415 times faster I actually mean 5.85% slower.
Expected that difference between M1 and M1 Pro to be bigger though, as PHPStan parallelizes using Paratest.
Used PHP 8.4, where you tested using PHP 8.3, will that differ much?
@OndrejMirtes@Ruud_@markusstaab@VotrubaT Howdy ππ» M1 2020 here, calling in with uncached `time make` results on same `PHPStan@cc4eb92` as the article.
2m39s
2m16s
2m16s
2m23s
2m30s
2m27s
Avg 2m25s.
So, the regular M1 seems 0.9415 faster as the M1 Pro baseline.
@markusstaab@VotrubaT Was considering moving from M1 to the M4 Pro, and this article convinced me to do so. Thanks a lot for the article, and also for including the Thinkpads in the equation.
@levelsio Such issues could be considered as good examples that Domain Events (e.g. "UserDidSignUp") on a simple user object, even without an ORM, could prevent. The event should get collected AS RESULT OF a successful database transaction, and then lead to infra / network related logic.