Re-creating a window after an inturrupt can crash and leak memory

[Brian-ED]

When interrupting code in Dyalog, it is very common to skip steps in execution or jump past cleanup code. This causes resource leaks and crashes.

[Brian-ED]

raylib very commonly segfaults and crashes if it's used wrong. raylib has a lot of assumptions about how it's used. Breaking these assumptions is very easy to do accidentally in the Dyalog session because of it being an interactive session where errors are supposed to happen, be caught, and iterated on.

One thing I considered but doesn't work is removing the resource-cleaning functions and instead have them activate automatically via destructors. The problem with destructors is that if a reference to some resource is assigned to an outer scope, it won't die, and these sorts of crashes will keep happening.

One realistic option is expanding the auto-error generation to also make assertions that do reference-counting-like cleanup automatically after closing the window. I haven't thought this through deep enough to know if it would work or not.

[arcfide]

I'm already seeing some potential performance problems with some of the error checking code in the way that we move from Raylib to APl and back, so I'm hesitant to add too much safety into the system because of performance concerns.

However, I do think that we've seen some surprise crashes on our end even when it seems like we have cleaned up the resources correctly. It's not exactly deterministic for us yet, but we'll keep an eye on it.

[Brian-ED]

Interesting, yes that's a bigger problem. Thanks for keeping an eye, making a reproducible example of those kinds of cases is difficult.