Sarkath
Active Member
- Sep 8, 2019
- 547
- 949
- 245
RAM limitations, I assume? OOM killers on mobile operating systems tend to be pretty aggressive.Thank you. I tried some ideas, one was simply by minimum edits and JavaFXPorts/gluon/... atrocity. Yet it lead me to discover that androidide would rather kill itself than give me a chance to test any instance of that code. Which reminded me that whatever it spits out will run even worse than it originally did on pc.
This, coupled with the questionable licensing terms and people reporting that Inno tends to reject significant refactors to the main codebase (I mean…I kinda get it?), is why I abandoned my efforts. I contributed one patch that significantly reduced save times, but I didn't even bother with some of my more experimental stuff (like swapping out the heavy JS engine with a much lighter expression parser).So my second option was partially optimising the existing code by relying on android specific features, but when you have such an entwined code you would have to rewrite a large but double chunk of it. Which also means that whenever Innoxia releases another update, larger parts could be rewrote or broken immediately. Which would still lag on android anyway, because the major bottleneck is in the background complexity.
Hah! Yeah, that checks out.So the final answer was to make a single build that's completely optimised for android, then it would never be updated again. Basically... Making the entire game from scratch! Well in hindsight...
I've established long ago that the best way to improve LT would be to build an engine from scratch that's compatible with its data files. The biggest issue with that plan is that when Inno made a bunch of aspects of the game data-driven she generally only used that for new additions, so all of the old data remained as Java classes. Pretty bizarre, honestly.