Not sure I follow, would prob have to see it to understand fully, to be honest. Having said that. My routine does not scan for bubbles that have more than one door. I would probably approach that type of bubbles differntly in that I'd start with a 2way and traverse it N steps until the max 'bubbles size' was exceeded.

Multi-door bubbles don't always start w/ 2 ways, but what I'm saying is this...

Door -> 2 way -> 3 way -> Door | 2 way -> Door

Or something to that effect. No dead ends anywhere, good bubble tho.

Sorry I'm so late to this thread, I've been hammering away on a haggle refresh and twx update, both coming along nicely by the way.

I have a nice "conventional" bubble script that I'll try to post to Grimy in the next few days, and I think it will be enlightening to you LS. It uses a well-designed avoid approach to locate bubbles of any size, including gold bubbles. Takes about 45 secs in a 5K, or roughly 10 minutes for a 20k. It is more of a brute force approach, made possible by setAvoid/clearAvoid and the getNearestWarps plotting in 2.04.

I'll refrain from talking about the more experimental techniques for the moment, as a good bubble algorithm discussion usually gets me distracted for weeks. I'll be glad to revisit this topic in a couple of weeks though. Until then, let me tantalize you with this: rev and I sketched an algorithm that takes twx 2 seconds for a 5k, but it does have it's limitations.

+EP+