AppCrash

[JakubChecinski]

Not 100% sure what caused it, maybe choosing not to normalize 3D vectors in a widget

The error message is attached as another image to: https://imgur.com/a/kHeqYxi

Edit: ok I think it was for 3D_CUBIC window actually, which could explain lack of memory error

[LemurPwned]

The crash is most likly due to the Text Selection bug - that also solves #124

[JakubChecinski]

I still get this error for 3D Cubic, even with normalization

The error message is almost identical but now also contains info about a runtime warning: https://imgur.com/j7olFzu

[LemurPwned]

Can you provide more info about the dataset and settings? Does it happen each time?

Runtime warnings are due to dividing by zero or very small number close to zero - these produce np.Nan values which allow later for not drawing an empty object

[JakubChecinski]

I checked different datasets and found an interesting thing: this error occurs only when loading full Examples->0520nm directory. If I load a single file or loading the Examples->0200nm directory, 3D Cubic widget works fine. Maybe my laptop really lacks available memory?

Settings: default 3D Cubic settings, I do not touch anything

Yes, for 0520nm directory it seems to happen every time.

[LemurPwned]

Does it happen for Arrow structures? This might be due to large memory usage -> 520 is a huge dataset. then each cube requries 72 veritces. Similarly, arrows also need plenty of veritces - 64 for each structure.

[JakubChecinski]

I tried to check for Arrow, but got bug #133 instead.

I also think it's memory, but can we provide a wrapper around it? I.e., instead of crashing the app would refuse to load a dataset that is too large and display warning? Or is it technically too difficult?

[LemurPwned]

(Same goes for #133 .) Wrapper is a straightforward solution. I might think about getting the data progresively loaded into ram as the animation progresses, however, there are two options:

• animation would pause/slow down while picking extra data
• data is loaded on the go but in a separate thread (more difficult, I am not sure if simple mutexing will be enough).

Anyway, how about when larger averaging or decimation is selected? The general problem here is that there is a time cap set for data loading, hence TimeoutError. For example, if there was a 100 files, they are all loaded in parallel and each has a time cap of 20 seconds by deafult to be processed. I can obviously increase the limit, but it seems to me that it won't fix an issue since 20 sec is quite plenty for a file to be loaded and processed.

Weirdly enough, it seems like a purely Windows issue. I have noticed that somehow processing takes much longer here.