Problem with Array Split and Build. I might cry.

[RJHollins]

just saw this.

I can confirm the original issue with FS 3.0.4

[Spogg]

Hey thanks for that RJ!

So it's not a new bug, it's a feature

Cheers

Spogg

[Spogg]

Hang on a minute...
The issue is there in 3.04, not there in 3.06 and there again in 3.08 and 3.09...?

This gets progressively weirder!

Cheers

Spogg

[tulamide]

Hang on a minute...
The issue is there in 3.04, not there in 3.06 and there again in 3.08 and 3.09...?

This gets progressively weirder!

Cheers

Spogg

No it's not. I just now worked with the schematic in 3.0.6 and it's the same behaviour. So either I was too tired yesterday to notice it, or I did something (without knowing) that prevented the issue. I assume the former.

Your fix and Walter's fix are basically the same. Your 1ms delay isn't actually 1ms, since the prim uses Windows timer, which is totally unreliable and can't even get lower than approx. 15 ms. Also, you send the trigger to the last splitter in chain. Notice the order of the green lines from the first array prim. At first a trigger goes to the upper section, that's completely executed. Then a trigger goes to the lower section, that again is completely executed. Then a trigger goes to the last splitter of the lower chain, which forces Flowstone to send backward triggers in order to update the input value. As a result of those backward triggers, the preset modules are executed a second time.

It is all about sending a second trigger, so that the preset modules are asked for their value again. As I said yesterday, this should not be the case. There's something in the preset handling that is messed up.

[Spogg]

Agreed

Our "fixes" aren't fixes at all, just work-arounds really. And Walter's behaves better in my real schematic. I just wanted to kinda prove it was a triggering thing, by using the method I did.

It's odd that this issue has been around for so long. I guess nobody made this topology before, so it's never been noticed. With near infinite possible configurations and complexity it's amazing that more of this sort of issue doesn't come up.

I wish Maik joined in with this stuff a bit more! I'll get in touch and see what he says...

Cheers

Spogg

BTW I do know that "near infinite" is nonsence

[RJHollins]

Agreed

BTW I do know that "near infinite" is nonsence

Oh ... then you've never seen one of my 'work' schematics. They are loaded with various ideas, routines. Most at one stage of failure or another.