I'd be interested to know if the rapid advancement of AI actually slowed down the development of XF — for the time being — so what they wrote / are writing could be double checked / bug free, and because of that, if we'll see fewer betas and RCs before a version release.
I'd be interested to know if the rapid advancement of AI actually slowed down the development of XF — for the time being — so what they wrote / are writing could be double checked / bug free, and because of that, if we'll see fewer betas and RCs before a version release.
I'm certainly finding my own recent development progress has slowed significantly while I use AI to help me build out my automated testing systems through extended unit testing; static analysis; container-based automated builds for testing specific configurations and php/platform combinations, etc.
It's fantastic to be able to tell my agent to go and build a XenForo site, install an old version of my addon, then test the upgrade process using the new version, while I continue working on something new. Or tell it to go and test various combinations of XenForo and PHP version - for example: let's see if my addon works as expected on XenForo 2.2.0 with PHP 8.1 and also on XenForo 2.3.13 on PHP 8.5, etc. This is all time-consuming work that often gets dropped because there's only so many hours in a day - but now it's an automatic part of our development process.
Note - I'm not just talking about CI work here - this is an agent actually building and exercising the install and doing various tasks to test that things are working in a much more sophisticated way than a CI runner would typically do, especially given it can diagnose issues and then go and propose a fix for them rather than just reporting success/failure.
I'm expecting that the future benefits of this work will be significant where turnaround times on testing new releases is dramatically faster, and I hope that my code - being much more thoroughly tested now - will also be more robust.
The move to modern PHP conventions has to be slow and considered by design.
Any change to an existing method’s arguments or return type would break any add-on that extends those methods due to a signature mismatch.
And the cost of that to most is much greater than any perceived benefit of code being annotated in a more modern way.
In fact I’d go as far as to say that what the code does or looks like or how modern it is just simply doesn’t matter. It’s us that has to maintain it, work within the constraints of older PHP and consider implications for all changes.
If you’re writing your own add-ons, you can use whatever code level you feel comfortable with. For new code that we write it always follows the most modern conventions our minimum requirements allow.
@Chris D The dear one, Could you confirm whether version 2.4 will be released by the end of 2026?
It is very important for me to know...
Thank you very much.
@Chris D The dear one, Could you confirm whether version 2.4 will be released by the end of 2026?
It is very important for me to know...
Thank you very much.
@Chris D The dear one, Could you confirm whether version 2.4 will be released by the end of 2026?
It is very important for me to know...
Thank you very much.