AA-fire/casualty selection issues in Revised (and other versions)


  • Admin

    @Panther If there is something ice-boxed, and it is an active issue and worth the attention, simply create a new bug. Ideally the new bug would be a very well written, short, concise, bug report with a link to the previous bug that was ice-boxed for more background and any other conversation threads.

    If you really don't want to go to that trouble, just copy/paste the closed bug report. That level of non-effort might not be that well received, but at the end of the day active issues should be tracked as open issues.

    AFAIK all items that are for bad game rules are still open and the issue expiry is being ignored for them. Those issues being on page 3, page 4 of the queue is not a good thing. It might feel good to have a ticket not be closed, but if it's still open and months old, it's not getting looked at. The time having passed is evidence that something has gone wrong, there is no reason to think that waiting N+1 months is going to lead to more success than waiting N months. Breaking up tasks into smaller tasks, making them more concise with easier to follow steps and a clear call-to-action of how to fix can be the difference sometimes.

    In general, I would agree that rules compliance is very important. Beyond problems that cause game crashes, correctness of the game rules is about more important than anything else I can think of.

    I think my ideal situation would be for the game to define 'rule sets'. In this you could select which rule set to play as an option. So for example you could switch to LHTR V3 rules as simply as you do when adding a bid or going to low luck. I'd like to see each rule under each rule set be selectable as a checkbox option. In this way every rule variant would find its way as a line item in a set of game options. On the code side, it would map to a pretty specific way the rules would be applied to the engine. Personally it's something I'd like to work on, but it's behind a few other initiatives. The barrier is that the current way the code is structured, it's pretty tangled and adding such options would not be clean, the rules are pretty hacked in and there is a lot missing. Due to the importance given to rules, that is why I'm ignoring any kind of ice-box policy for those issues even if them hanging out at the back of the queue is effectively the same thing.


  • Admin

    Re: ice-box, one item I did not mention is just prioritization. Time elapsed is a proxy for importance and priority. I don't ever want to call a problem low priority, but some impacts are more significant than others. I'll also note that nobody had more issues get ice-boxed then me. I think I had 30 or 40 some problem reports get closed due to age.

    Again though, if something is an active issue and there is no open bug, just file the bug report. Part of the goal is we drew a line and then said "anything we get we'll fix within 6 months, or we'll be honest and admit we'll never fix it". Then you try really hard to make sure things get fixed within those 6 months. If an item is worth re-adding, it's important, and it'll then (ideally) get fixed.

    One problem of course is our capacity is very low, overhead is really high. So as much as I'd like to say I'll fix all these problems, the reality is I'm going to get 5% done of what I want towards this project over lifetime. It's one reason I keep bitching about maintainer efficiency, in some scenarios I see us spending a lot of effort to not even get 5%, but instead just 1%. I'll be really happy if a third of what I want to get done, gets done in the next half year or year.


  • Moderators

    @Panther said in [Open] AA revised bug (AA-fire/casualty selection issues):

    @Cernel said in [Open] AA revised bug (AA-fire/casualty selection issues):

    On this matter @Panther did we fully officially clarify casualties selection for v5/Global too (where you have AA factories and AA guns for battle only, with limited shots (and escorts/interceptors in case of Global))?

    We have analyzed that during this thread and identfied "only" the randomness of casualty selection as issue.

    I'm not sure what you mean. I believe v5/Global are not bugged, as in those games there is not any kind of randomization (or groups separation) of AA casualties.

    However, at the GitHub issue, I see that, for v5/Global, you have listed "attacker cannot choose casualties (casualty apparently is taken out on a random basis), tested with factory-built-in AA gun against a bunch of bombers)".

    I've tested "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition" a few times using TripleA 1.9.0.0.13066. In a total of 4 tests for each one (for Global2, I had the units never starting from a territory with an airfield) in which any bombers were shot down, the program always took out the ones with the lowest movement left; so, aside from this being a coincidence, it doesn't seem to behave purely random like, say, v3. Moreover, I tested bombing raiding with both bombers and tactical_bombers in Global2 (editing the factory and the fighter out of France, then sending all bombers and tactical_bombers at reach to raid the airfield), and it allowed me to select AA casualties between tactical_bombers and bombers (I got 1 hit and I was able to select either a tactical_bomber or a bomber as casualty), as I believe it should be. I cannot do this second test for v5, and I know the only way to test this for sure (beside reading the assigned properties in the xml and assuming they behave coherently, at least) would be making a mod in which we have multiple types of bombing raiders (like giving bombing ability to the fighters too), and see if the program lets you pick the type. Anyways, I can at least say that, based on what I know, I cannot confirm there is a bug here, until it happens to me too. If you want to upload a savegame showing in v5/Global the behaviour that you have described in GitHub, I'll take a look at it.

    On this matter, but more in general, when you can fully select casualties (that is the case for AA guns too in v1 and v5/Global, but not for the rulesets between those), TripleA always takes (amongst the same type) aircrafts with the least remaining movement first. Am I correct that, for all games since v1 onwards, this is arguably a TripleA issue, as, instead, you should be able to take aircraft with higher movement left first, if you so want (and TripleA doesn't allow you to)?

    So, to be exact, for v5/Global only:

    1: The AA Guns are not bugged (beside the fact that aircrafts with the lowest movement left are always selected first), and work as intended, as they are supposed to work just like Classic, except only for battle and with a 3 shots each limit.

    2: The self-defending factories under bombing raids are not bugged (beside the fact that aircrafts with the lowest movement left are always selected first (but in case of bombers, this should be actually what you always want)), and work as intended, as they fire only at bombers (never at fighters), and the defender is supposed to freely pick casualties, anyways.


  • Moderators Admin

    @Cernel said in [Open] AA revised bug (AA-fire/casualty selection issues):

    Anyways, I can at least say that, based on what I know, I cannot confirm there is a bug here, until it happens to me too. If you want to upload a savegame showing in v5/Global the behaviour that you have described in GitHub, I'll take a look at it.

    I have uploaded that savegame already here:
    https://forums.triplea-game.org/post/16633

    Another user experienced it as mentioned here: https://github.com/triplea-game/triplea/issues/4133#issuecomment-445928137


  • Moderators

    @Panther Any information on the TripleA version used to make those savegames?


  • Moderators Admin

    @Cernel You can use the current stable to look into it.


  • Moderators

    @Panther Using the current stable, I confirm it happens to me too on the game "World War II v5 1942 SE TR". However, unless that is a stupendous coincidence, "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition" appear always taking the bombers with the lowest movement left first.

    Can you test the same situation on those games too, se if it happens there too (you don't need to send any fighters in attack; I've no idea why you did). I'm starting thinking it might be something wrong with the "World War II v5 1942 SE TR" mod only, assuming this is not related to some changes for that "tournament rules" version only.


  • Moderators Admin

    @Cernel said in [Open] AA revised bug (AA-fire/casualty selection issues):

    Can you test the same situation on those games too, se if it happens there too (you don't need to send any fighters in attack; I've no idea why you did). I'm starting thinking it might be something wrong with the "World War II v5 1942 SE TR" mod only, assuming this is not related to some changes for that "tournament rules" version only.

    I used fighters to test any possible casualty selection from the regular AA fire, which did not occur in that scenario.

    Apart from a slightly different setup ( http://www.harrisgamedesign.com/phpBB3/viewtopic.php?f=43&t=18945&start=8#p67254 ) the game as well as the xml is identical.


  • Moderators

    @Panther So, you confirm you are unable to reproduce this bug on "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition"? If so, I guess I'll take a look at the properties for "World War II v5 1942 SE TR" and at those tournament rules; see what's up with any of them.


  • Moderators Admin

    @Cernel said in [Open] AA revised bug (AA-fire/casualty selection issues):

    @Panther So, you confirm you are unable to reproduce this bug on "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition"?

    How do you come to say that? It was pure coincidence that I took the TR version instead of the other.
    Again it is only the setup that is slightly different.


  • Moderators

    @Panther What I asked you is to do the same test you did on "World War II v5 1942 SE TR", but using "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition", instead (you don't have to send any fighters, just the bombers).


  • Moderators Admin

    @LaFayette Thank you for your comments that I understand very well. I just wanted to express my sentiments when the result of all the work put in bug-hunting and investigation might be "ice-boxed".
    😉


  • Moderators

    @Panther said in [Open] AA revised bug (AA-fire/casualty selection issues):

    @LaFayette Thank you for your comments that I understand very well. I just wanted to express my sentiments when the result of all the work put in bug-hunting and investigation might be "ice-boxed".
    😉

    My only issue is really that I think that is going to keep butt-hurting new people, in the moment they see the issue they spent a serious bunch of time sorting out gets just closed after some months, likely discouraging people from bug reporting anymore. Myself, I start seeing that there is not much difference between an issue getting forgotten at the end of the list or closed. I don't think I'm going keep copy pasting the same thing at 6 months intervals, either.


  • Moderators Admin

    @Cernel said in [Open] AA revised bug (AA-fire/casualty selection issues):

    @Panther What I asked you is to do the same test you did on "World War II v5 1942 SE TR", but using "World War II 1942 Second Edition" and "World War II Global 1940 2nd Edition", instead (you don't have to send any fighters, just the bombers).

    First try:
    Global: AA scored one hit, engine removed the bomber with 2 movement points left:
    GlobalAAtest_ncm.tsvg


  • Moderators

    @Panther Ok, wow, looks like. I did a test like that and for 4 times in that one and 4 times in v5 and it took out the ones with the lowest movement left in all cases. Can you confirm you are using TripleA 1.9.0.0.13066? I guess I'll do your exactly same test too, now.


  • Moderators Admin

    @Cernel Confirmed.


  • Moderators Admin

    @Cernel Just for completeness the v5-OOB-savegame, first try, again one aa-hit and again the engine removed the bomber with 2 movement points left.
    v5test_3ncm.tsvg


  • Moderators

    @Panther Did it ever happen that, when you got hit, the bomber with 2 movement points left didn't die?...

    I think there is something more going on here than purely a wrong setting (as also you can see the property "Choose AA Casualties" is enabled, differently from v3 etc.).

    So I believe this is yet another bug, not just v5 working as v2 etc. (bugged too, but in a different way).

    I can tell you, doing something like in your save, every single time the AA scores 1 hit (or more) with that save, the bomber with 2 movement left die. With randomness you can never be sure, but I'm starting doubting we have a randomness problem here...

    What I'm almost sure it is going on here, actually, is that this is not even a case of using a pure randomness casualty, but, instead, taking out fighters with less movement left first, except for skipping the one with only 1 movement left. So, if 1 is lost, it will be the one with 2 movements left. If 2 are lost, they will be the one with 2 movement left and the one with 3 movement left. If 3 are lost, they will be the one with 2 movement left, the one with 3 movement left and the one with 4 movement left. And so on. Likewise, if you have 4 with 3 remaining movement points and 1 with 1, and 2 are shot down, I think you will always end up with 2 with 3 MP and 1 with 1 MP surviving.

    On top of this, this skippage of the bomber with 1 movement left happens only in some cases, as I'm sure I tested other cases a lot before, and this was not happening (all apparently worked properly).

    p.s.: Can you confirm that, when not random (like the battle AA of v5), the fact that TripleA takes out the air with the lowest movement left is wrong, anyways, as you should be allowed to take out air with more movement left, instead, if you so wish?


  • Moderators

    @Cernel I've run this same situation so many times, when 1 hit is scored, always the very unlucky bomber with 2 movements left is killed. The ones with 1, 3, 4 and 5 movement left always survive.
    You can try doing it yourselves, again and again: just load this save and click on "Done":
    globalaatest_ncm(Cernel01).tsvg


  • Moderators

    @Panther said in [Open] AA revised bug (AA-fire/casualty selection issues):

    Apart from a slightly different setup ( http://www.harrisgamedesign.com/phpBB3/viewtopic.php?f=43&t=18945&start=8#p67254 ) the game as well as the xml is identical.

    I think I would have done that as a custom option of v5, and a set of triggers that makes those setup changes when enabled. After all, this is more or less the same deal as the bid settings.

    Or really just change v5 to that and that's it, especially in the moment this game is really far from popular, at least in lobby. This is what we have done for v6, just using the revised setup, instead of the incredibly broken OOB one.


Log in to reply