INI Crash and Slowdown

Hey,

So i ran into a werid behaviour today. I use INI to retrive different languages, however that have slowed down level selecting that is using array object.
While playing around with INI I managed to get a consistant crash. If i delete INI object and then try to read from it, it crashes runtime, example included.

When it comes to lag, I dont really understand why it appears as I only load ini at the start of frame and that is it. Its fine for all of the locations besides level selector frame where i load in preview via array. Sadly i cant post this version here, so if anyone from clickteam wants to look deeper into that issue please let me know.

Video is more or less here:
https://upload.catgc.com/img/a72b986...020d2d2496.mp4

Edit: I found somewhat of a workaround, I can point towards a different ini that does not exist and it stops lagging. But that behaviour sounds still out of place for me.

I see what happens, a bit complicated to explain. As the INI object doesn't exist, the GroupItemString function (line #3) doesn't get completely evaluated (= it stops before reading all the parameters) and this crashes the expression evaluator. I don't think this can be easily fixed. Why do you destroy the INI object? You shouldn't need to do that.

Quote:

---

When it comes to lag, I dont really understand why it appears as I only load ini at the start of frame and that is it.

---

The INI object uses the Windows INI API by default, i.e. I think it may access the file each time to read a value. If your INI is not updated by another application or sub-application, then you should unselect the new option "[Windows] Read changes from disk" that comes in the 2.5+ DLC. When this option is unselected, the INI file is loaded in the Set Current File function and the values are read from memory.

Thanks for such a quick response, the crash does not really bother me, its just something i found out. However what bothers me is following:

-Start of Frame
-Length of ini string is longer then X
--Change string to ini value

Then I would assume that after the first "Start of Frame" condition, the line would fail and further condition would not even be looked at. That would be good optimization vise, but I understand that there are some cases where it would make sense to check further events even after first condition fails. Maybe a new addition to "limited condition" where after first condition is failed it would skip the whole line, it would be really useful for optimization.

Still lags even with "[Windows] Read changes from disk".

Edit3:I guess workaround would be to disable the whole group of events after certain period of time.

sorry..double post.

Quote:

---

Originally Posted by wii1mii

However what bothers me is following:

-Start of Frame
-Length of ini string is longer then X
--Change string to ini value

Then I would assume that after the first "Start of Frame" condition, the line would fail and further condition would not even be looked at.

---

I think that's exactly what happens. After the start of frame, the 2nd condition gets ignored.


If I do this: then hitting the delete button will cause a crash:

http://bit.ly/2YhSZSh


But if I do this: then there's no longer a crash, which to me suggests that Fusion's no longer reading the 2nd condition:

http://bit.ly/2YlUSNK

Quote:

---

I think that's exactly what happens. After the start of frame, the 2nd condition gets ignored.

---

It would make sense yes, but then I completely do not understand why following
https://upload.catgc.com/img/c98970f...8540666943.png
condition causes lag. As "LangSwitch" loop does not even happen on particular frame.

Basically have about 120 conditions that look really similar to above picture. And they causes lag that you can see in the video i posted earlier.
If i disable that group of events it is fine.

What happens when you split the event in the video into two events: one event for start of frame, and one for on loop? Do they both still cause lag, or only one of them?


By the way, why do you need check for start of frame OR on loop? Why can't you just check for on loop, and in another event, run that loop on start of frame?

Quote:

---

What happens when you split the event in the video into two events: one event for start of frame, and one for on loop? Do they both still cause lag, or only one of them?

---

Basically as soon as i point ini object towards the ini file it starts to lag, no matter if its start of frame or later during loop.

Quote:

---

By the way, why do you need check for start of frame OR on loop? Why can't you just check for on loop, and in another event, run that loop on start of frame?

---

Well yes i assume i could optimize it by removing start of frame with or and just have it done via loop. Gonna try that now and revert back with update soon via edit.

Edit: seems like that actually fixed the issue. Interesting. Well thanks for pointing me in the right direction!

Glad you got it fixed! But I'm now quite curious about what the issue was. I wonder if it's a problem with Start of Frame, or fastloops, or "OR", or something else :o If I had to guess, I think I'd guess that the problem might be to do with putting a green condition (eg. "on loop") underneath OR (where it no longer gets displayed as green).

Quote:

---

Originally Posted by wii1mii

Basically as soon as i point ini object towards the ini file it starts to lag, no matter if its start of frame or later during loop.

---

So, just to clarify, when you changed this event:

http://bit.ly/2YhmP9k

into these 2 events:

http://bit.ly/2Ymx643



You got lag in all three of the following situations, yes?

http://bit.ly/2Ymx643

http://bit.ly/2Yn5YlI

http://bit.ly/2YmxdfZ




And if you don't mind, could you try replacing OR (filtered) with OR (logical) in this event and seeing if you still get lag?

http://bit.ly/2YHGeLT


EDIT: I wonder if the problem is related to having two green conditions, because I just noticed something. If you replace Start of Frame with a non-green condition, then the On Loop condition suddenly becomes green

http://bit.ly/2YIb8Eh


If you do something like that, does the lag go away? If so, then maybe it was having the two green conditions in one event that was causing the issue (even though, logically, one would think that it doesn't matter, since you're not combining the two conditions but rather testing them separately with OR).

Issue was rather not in:

https://upload.catgc.com/img/3bcc8cf...af77379fbb.png

But in:

https://upload.catgc.com/img/8b9fcc5...e450b99e4d.png

Since as soon as it activated everything started to lag with previous condition set

https://upload.catgc.com/img/c98970f...8540666943.png

With OR (logical) also seems fine.

https://upload.catgc.com/img/af02634...fdacbbabdf.png

I guess that narrows it down to OR actually making fusion to check for every condition in the line, I guess that makes sense now that i think about it.

I'm confused. Here you say that you think it's an action that's causing the problem

Quote:

---

Originally Posted by wii1mii

Issue was rather not in:

https://upload.catgc.com/img/3bcc8cf...af77379fbb.png

But in:

https://upload.catgc.com/img/8b9fcc5...e450b99e4d.png

---

But here you seem to suggest that you think it's a condition that causes the problem.

Quote:

---

Originally Posted by wii1mii

https://upload.catgc.com/img/af02634...fdacbbabdf.png

I guess that narrows it down to OR actually making fusion to check for every condition in the line, I guess that makes sense now that i think about it.

---

To me, it seems like the problem is about a condition rather than an action. Even if it's an action that causes the lag, the mystery is - as you said earlier - why certain conditions are allowing that action to trigger at all.



Do you get the slow down if you replace this

http://bit.ly/2YoG94u


with this?

http://bit.ly/2YIb8Eh

From my understanding:

When ini is set to an actual file then the condition actually loads the file to check if the length matches.
Loading file so many times causes lag when something rather heavy is performed.

Now no matter about the beginning condition, as soon as the ini file assign action is performed everything slows down.

However once i removed OR performance was back to what it used to be.

So thats why i suggest that the OR forces fusion to actually check every condition in the line.

--
I mean i tried to change
https://upload.catgc.com/img/3bcc8cf...af77379fbb.png
To "On button press" and as soon as i pressed button it started laging, so has to be action.

But why does http://bit.ly/2YpaBvq cause lag but "on button press" doesn't?

I'm thinking that the problem might be with the "on loop LangSwitch", which is not green (but it should be!). I wonder if it's being somehow ignored and - since it's been ignored - effectively treated as an "always" condition, which makes the action trigger each frame. What happens if you replace Start of Frame with a black condition that returns false (which would make On loop LangSwitch become green as it should be)? Does it still cause lag? That's why I'm asking you to try this :

http://bit.ly/2YoN3Xs

Here,

I hope this makes some more sense:
https://upload.catgc.com/img/7f764e4...e4b1cb1c74.mp4

So it lags when you press control? Well that's as expected. But the mystery is why it also lags when it has an "on loop" condition even when you don't call the loop. What happens when you replace start of frame with a black condition, as above? Does the action still get activated?

https://upload.catgc.com/img/a9224b8...2e05cec3e2.mp4
Even with first even being black it still lags yes.

ok, thanks for humouring me and checking that. I thought that maybe the problem was that when "on loop" was black instead of green, it ceased working properly. But that doesn't seem to be the problem.

So actions that are not actually played can still slow down the runtime. Weird. Actually, this reminds me of something casleziro wrote in another thread recently. I guess it might be connected, since both are unplayed actions that cause Fusion to load stuff:

Quote:

---

Originally Posted by casleziro

I recall Nifflas using "Never" to "load sounds" into Fusion memory at start in one of his sound engines. It was just a "Never" condition with a bunch of "play sound x" events in it. I'm not positive, but I assume Fusion will load these sounds since they're referenced, even if they are never played? He had a comment that said the line was to do just that to avoid a small lag the first time a sound is played.

---

Well like i mentioned, there are few workarounds.

a) "Start of frame"/"On loop" [enable group] to load in the ini and go though every event and at the end [disable group].
b) Remove "OR" from the condition and only use loop (currently using that)

Theoretically could combine both but i dont see a reason for that.

c) Use loop to point towards INI file and at the end point towards file that does not exist. This is a dumb work around but theoretically it works.

None the less, thanks for being here and helping me out ;)