Discussion: p5 v1 timeline

[quinton-ashley]

Topic

What's the timeline for p5 v1's development?

Given the release of p5 v2, I'd prefer for p5 v1 to be preserved as is, since even bug fixes can have unintended consequences and maintenance costs.

I suggest that only critical LTS updates be made to p5 v1 and all further development be focused on p5 v2.

---

I'm asking this because recently I've been struggling to maintain p5play's compatibility with p5 v1. Several updates to v1.11 have broken stuff that p5play was relying on, causing games to crash or not even load.

Most of it is my fault for depending on internal state like _setupDone and overriding methods like loadImage, which p5 warns against. Yet bugs like #7819 were also introduced. It's caused a lot of confusion and annoyance for p5play users.

For example this game made in 2023 currently can't load with p5 v1.11.6 and above.

https://pannawit0.github.io/SkullKnight/

That's cause p5play can no longer add a prop that store's the image's url to p5.Image objects. I also can't think of another way to do it if #7928 isn't implemented.

I didn't expect such internal changes to occur so late in p5 v1's development.

[ksen0]

Hi @quinton-ashley ! The intent is indeed for p5 v1 to be preserved. The previous intended timeline we shared #7488 proposed:

1. There are currently no feature additions on p5.js 1.x, only improvements to documentation or bugfixes, until at least March 2026
2. After that point (March 2026) there would also be no further bugfixes

"Several updates to v1.11 have broken stuff that p5play was relying on, causing games to crash or not even load" - are there regression tests that can be added to 1.x to prevent this? Perhaps covering what you rely on that might not be covered in the suite?

Re: 1.11.6 and the url issue: I was looking at the release notes and I'm struggling to see what would have that effect; it is documentation updates and an update for global state that seems you reviewed. Or did it happen earlier?

Here's an example of a bug #7818 that as far as I can tell, was not introduced in 1.11 (same behavior in 1.0.0: https://editor.p5js.org/ksen0/sketches/194QwScAV) - Do you think things like that should be fixed?

[quinton-ashley]

The intent is indeed for p5 v1 to be preserved.

That's great to hear.

are there regression tests that can be added to 1.x to prevent this?

Well I think there shouldn't need to be tests for stuff that isn't part of the public facing API. idk

It is strange that when I switch from v1.11.5 to v1.11.6 p5play can't extend loadImage anymore.

/*! p5.js v1.11.5 April 17, 2025 */

Top of v1.11.6 says it's v1.11.5 but the date is different.

/*! p5.js v1.11.5 May 14, 2025 */

Not sure what the cause is.

[quinton-ashley]

#7818 should be fixed?

Nah it's fine. Just closed that issue and explained the actual problem better (my example was bad), but that was a problem introduced in v1.11.7.

[davepagurek]

Top of v1.11.6 says it's v1.11.5 but the date is different.

When a build is made, it grabs the current date and the version from the package.json, so this probably means we accidentally made the new build before bumping the version number when doing the release. So if the date is different, it looks like this is the correct build. Something for us to look out for when doing the next build I guess!

How is the overloading of loadImage done?

Looking at the 1.11.5..1.11.6 diff, some initialization ordering changes, which could be related depending on what triggers the overloading.

[quinton-ashley]

@davepagurek p5play extends loadImage like this:

let _loadImage = this.loadImage;

this.loadImage = function (url) {
  let img = _loadImage.call(this, ...arguments);
  // ...
  img.src = url;
  return img;
};

[davepagurek]

I tried playing around here: https://editor.p5js.org/davepagurek/sketches/DN6P4q46Y

I think any JavaScript class method, when called as a function, needs to use .apply or .call in order to make sure this is mapped correctly. Otherwise, this will be undefined and may cause internal things to break. Does that fix p5play's usage?

[quinton-ashley]

Nah I just forgot to write that in the code excerpt I shared.

The crux of the issue is that for several years p5play was able to override this.loadImage in the "init" addon method hook and now suddenly since v1.11.6 it can't.

But yes, you're right that I could override p5.prototype.loadImage as a workaround.

Ideally I don't want to have to make any changes to p5play to maintain compatibility with p5 v1.

For now I'll just recommend that p5play users stick with v1.11.4.