Bots and Simulations > Bot Tavern
The F1 League lives again!
EricL:
--- Quote from: bacillus ---I wonder if the top bots in the new versions would do as well in the old versions, where the dinosaurs ruled, because some of these bots (especially Aura and "simple bot virus") are unfathomably incompetent.
--- End quote ---
Many would not as they rely on features only available in newer versions such as eye width or .totalmyspecies.
As always, I would be interested in cases where older bots have problems in newer versions as a result of unintended changes. My goal is to provide as much backward compatability as possible. Sometimes we break from the past and intentionally change things (such as making eye sight distance a function of width) but that is pretty rare. If an old bot doesn't work as it should in newer versions because of some unintended change, I want to know about it.
Moonfisher:
I think Aura fell appart because of the eye range, generaly any old bots using 360 degree eyes are having problems.
And I think the reson Excalibur fell appart recently is because of the reactive armor using in/out, it seems like it's making too much shell and poison and such...
Although it shouldn't matter since it's using the defined names and not the adctual adress... maybe one of the locations used has been put to use in newer drops...
Another good reason to use those free locations between 50 and... 175 I think...
Anyway a lot of the bots should be possible to fix without too much effort, atleast if they defined all their vars...
This is also why I've started to use constants and vars for everything in my new bots, it makes it easyer to understand your own code without spamming comments, easy to change keys or tweek magic values or correct the locations used by vars if they're no longer available... and this way if it breaks down in a new drop it should be possible to fix only by adjusting a few values.
EricL:
Sorry to continue off-topic, but one potentially fixable source of problems for older bots may well be the change in the eye value function that accompanied making eye sight distance a function of width. See this topic. I changed the function that determines eye values as well as making it based on edge to edge distance instead of center to center distance. This was a huge change, one I'm surprised has not ellicited more feedback. The specific details of the new formula are in the later posts in that topic. These changes impact eye values (I.e. what value corropsonds to what distance) for all bots, even those that do not change their eye width. I tried to choose a formula that was close to the old one for distances most bots care about, but it is quite possible that older bots that trigger at specific eye values to shoot, chase, etc. are not working as they were designed due to changes in the eye values for specific distances.
I am very much solicting feedback on this. I would very much apprecaite it if someone could perform some empirical compartive tests on eye values relative to actual physcial distance, say between version 2.43 and any recent version post 2.43z.
bacillus:
The 360 degree eye has nothing to do with it in most cases. Run an Aura for yourself; even if you have 1000 veggies in a minimal-sized map, they die a repro death.
Numsgil:
It could be the eye values. I tried fiddling with them in the program once a few years ago and it generally caused some really nasty problems. Hard part is figuring out a way to test it.
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version