Unlike my application of the great detective's method during the development of Doom-3-Bravo, in the case of Legend-9 - most of my deductions turned out to be incorrect. I have lined up over a dozen of possible suspects to the problematic issues, and except one - they all had nothing to do with it. And even that single one, - was only responsible for one specific glitch, while everything else remained a complete mystery. However, i still managed to arrive at some new conclusions during my intense experiments over the last three days. That does come with good news and bad news.
Good news first: i have made 4 complete no-cheat playthroughs of the first hub, and another two prior to that - by using various cheats. During those first two runs - i have discovered many small issues that amassed to a formidable number at the end. I have addressed all of them, making multiple changes to all the 1st hub levels to improve optimization, eliminate overlooked level lapses, reduce reliance on triggers while restructuring some of the key triggers, updating balance and even removing certain fragments withing levels that were deemed unnecessary or not contributing enough. While i cannot claim that whatever i did - helps the problem directly, but in summary it allowed Doomsday to handle things better, thus reducing a chance of a crash.
During the four no-cheat test runs, i was able to progress through all the levels, moving back and forth without any level-related crashes whatsoever. Yes, since i used all characters, - even the ones that i claim to have weapon-based crashes, - i have discovered something else there too. First of all, Elf class did not crash even once through the entire run. Mage class crashed once. So i can only guess that none of the potential crashes have anything to do with the levels or with weapons. It all hinges on something that Doomsday cannot handle, or cannot handle when that "something" becomes too numerous or whatever. And i have no idea what that "something" is. And that's the bad news.
I had plenty of ideas regarding what could potentially cause a crash. And they all turned out to be wrong. I have three things to take care of in the correction of this build, and i want another test run, just to be sure. Then i'll be ready to upload the mod in a fixed version. There are plenty of smaller changes, and they make quite an impact. I will post a complete list of what's been fixed. I reckon by 23.10.2023 the fixed version will be up.
Glad that it's still being fixed since it came out few days ago. I could have played it as soon it releases but I was pretty distracted with other games with a few friends of mine and working on my Wolfenstein mod lately.
Besides that, I'll give it a try when it gets patched.
Yeah, that underlines the importance of longplay testing, and not just incremental or selective testing. No matter how much you do of the latter, - there is no getting away from the former. The only question is time, as in when.
That build has its fare share of issues, but i intend to launch the updated one tomorrow.