Chris21wen wrote: Thu May 23, 2024 5:58 am
A few questions first:
PBEM or AI game?
AI
The 2 other ports circled in blue (Bihoro and Kushiro I believe), are there any TF's "based" from either?
No
Game date?
Sept 42
Scen?
#1
About how many turns ago did you first notice this?
About 4 months, then 2 then 2 days ago.
Thanks!
Chris;
The supply/oil/resource movement code has always appeared to be the most fragile to me in the game. Other than breaking the pilot limit in the game, this portion causes me more challenges to finishing games than any other item (excepting of course my "stupids"). Anyway, remember my disclaimer:
I am not a dev, have never seen the code, and I do not recall the devs ever making many comments on this portion of the game.
Having gotten that out of the way, I can deduce a fair amount from having used the code for so many years now. Clearly, one of the first steps in the process is that it identifies boundaries in terms of where these resources can flow, then it will look at base sizes to get frequency and amount of movement and then finally push around. If there is an error in the first step, nevermind how, then the whole process falls apart. Then realize how large of a data pool that this game represents, huge. Just the "map" itself is huge in terms of data. Each hex has a number of properties, one part of which is hex side properties. Hex sides themselves have a large number of properties defining ownership, movement, bordering hex, etc. Now imagine just one value gets corrupted in all of this ... how? just call it data leak for want of a better term.
How to prove it and how to fix it are the same solution; go back to the last turn where it is working and re-run the turns from there. IF you get a different result, then you just proved a data leak was the likely culprit. IF you get the same result, then one of 2 things: either you need to go back further because the error is in your turn already OR we have a bug to report meaning that something you did (a movement, a battle outcome, something in the game) caused a hex property to NOT update correctly.
The difference between these two errors is reproducibility. If you can re-produce it, you have a bug. Catch is, that is true only if the code is not already corrupted. And the only way to know that is to keep going back until it doesn't appear and then rerun. A Dev might catch it easier/faster via a checksum type of algorithm, but we players do not have that in our toolbox, so we have to do the hard way.
Now we both know that the second option is not likely simply because of the millions of game hours played already. Possible, but not likely. Thus you are left with the first one: data leak. How to combat it? Save after every turn exit the game and restart WitPAE.
In every game I have played the last 10 or so years, I probably come across a data leak about every 100 or so turns. I can't remember the last time I had an issue EXCEPT when I break my rule about save/restart. Almost every time I have an issue, I review my turn notes and find that I have no mention of a restart ... When I break my rule, I kick myself because then I might have to go back 10, 20 or more turns before I get to my last "clean" turn and can commence running again. I routinely run my games to '45 and later. My record in terms of longest game is April 17 1946 ... disappointing it took the IJ that long to own the map, but I never said I was a good player ... I just have a lot of fun along the way., especially late game with all the cool toys that both sides get.
My final point is this: Can you get +1500 turns in this game? YES! Do you need to exercise some care getting there? Also Yes.
