wthit56

Heads up: that first jpeg link works fine for me.

Sounds like it's basically reminding the AI of some data each time a prompt is sent to it. What I do is just put it in the prompt, and it seems to work fine.

Those generators you listed all do the same, reminding the AI of things it needs to know about, but are pretty complex and have a lot to send, most likely. You can make your own Text AI pretty easily, and send whatever you like with the prompt. Maybe try that.

Thanks for the update! I get it. It's a lot of work, and a balancing act for timing things as well. It's hard to live up to (a minor group of) people's expectations. But knowing what the state of play is with an update like this really helps 👍

It's difficult. Because it's generally trying to make one image, building from one idea.

You can try using the BREAK keyword to separate groups of meaning. Like blue hat BREAK yellow dress is more likely to not mix up the colours for the skirt and hat.

The key is, $output is a function. With properties set on that function. So, that requires some sort of build to happen when you first access $output. Your test doesn't have any of that, which is why it works ;p

Here's an example of properties not working correctly on an $output function: https://perchance.org/sqfppqoq6e#edit I did try various tweaks and workarounds. Got something working within the page (can't remember how that worked now) that didn't work when imported. But this example doesn't even work within the page.

In another comment on this post I talked about my own expectations.

And that example I linked to in the post has some very weird behaviour being demonstrated.

Okay understood. Thanks for looking into this for me 👍

Yep that looks good 👍

Thanks for the test. That's what I guessed was happening, yeah... cleared first, processed, then appended. Most likely none of this will change, as it's kinda core and changing it could blow things up. But, for science...

I guess first the expectation/spec should be defined of what "should" happen. This is (or was) my expectation:

1. Square brackets define a function (the code within the square brackets) and a spot in the HTML.
2. The function is run.
3. Its result replaces the spot where the square bracket text used to be.
4. Update() will do the same process: run the function, and the same spot is replaced with the new result.

(There are of course other ways of defining things. I just don't think it's actually defined by the dev anywhere, but more cludged together--as is the style of perchance. 😉)

So the tricky part here is, in step 3, what happens if the spot no longer exists by the time the function is processed the first time? My expectation:

• The spot would not be replaced.
• And if the square brackets don't exist by the time square block processing gets to it the first time, it would not be processed at all and could not be updated--something like that.

If you rewrite the innerHTML/innerText, the old text nodes are disconnected from the DOM. So then you can check if the old ones still have a parent (or the same parent perhaps). If not, don't do the replace step.

Similarly for an update, you could keep a reference to each top-level element created by the code block. Then in the replace step remove as many of them as are still within that same parent, and add the new stuff in at that point--"replacing" it with the result.

I would also expect that if the entire contents of an element is square brackets, the entire contents would be lost on an update. So that would be an easier edge case to process in any case.

I just updated with a test5, which has a more complex case. The result is pretty wild! 😂 I don't know how anyone would've guessed that is the expected behaviour; seems pretty complex. 🫠

I know the assignment would return the string also. But then you'd think it would be the same as:

var result_from_square = (test.innerText = "text");
elem.innerHTML = result_from_square;

Which should be doing the same exact thing twice. But it's not, hence the post.

There are other oddities around this as well. Adding Text nodes to the element, and then updating the element which should wipe it... leaves those text nodes there. Stuff like that. It's all likely from the specifics of how it updates and runs square code. And most of the time won't matter. But when it does matter, it's confusing, and needs more and more specialised knowledge about the innards of perchance to even understand what's going on and how to fix it.

@[email protected]

You can use localStorage to save and restore things with code. I believe that's what the remember plugin uses to save things anyway.

So you'd do something like localStorage.setItem("last generated", data.generatedText) which will save it. And when you load, you can use loaded = localStorage.getItem("last generated"); if (loaded) { element.innerText = loaded; } which would put the text in an element with the id "element", if there was anything stored in the first place.

You can look up more about localStorage on MDN.