AA-fire/casualty selection issues in Revised (and other versions)
-
I would also suggest keeping a property or something for allowing having truly random casualties selection, even tho that doesn't actually apply to any basic games.
-
@panther said in AA revised minor bug:
@cernel said in AA revised minor bug:
Of course, in solving this, it would be necessary to take care not to bug v3 off, if you can also confirm that in v3, instead, as a departure from LHTR, you target the specific unit, thus possibly taking out same aircraft types with higher movement left first (albeit also the v3 rulebook is a bit confusing on this point, as it affirms that there is no need to differentiate hits if all aircrafts are of the same type, without mentioning the case of same type but with different movement left; so I'm not 100% sure on this point, either, but almost so).
Actually selecting casualties in v3 is not different. You differentiate for example fighters from bombers but don't differentiate if all aircrafts are of the same type.
In this context a statement of Krieghund (from 2009) came to my mind:
See https://www.axisandallies.org/forums/index.php?topic=12601.msg456162#msg456162We can read here, that "the authorities" have been aware that a (material) differentiation by movement points is lost because of grouping.
In any case, are you sure you can absolutely confirm that v3 AA casualties selection is exactly the same as LHTR (mandatorily grouping by type).
Asking because the v3 rulebook is definitely not as clear on this point as the LHTR clarifications are, and I really wonder why is that, since it comes after those, and this clarification you linked:
https://www.axisandallies.org/forums/index.php?topic=12601.msg456162#msg456162The intent is that each air unit is fired upon individually. However, the rules allow for all fighters to be rolled for together and all bombers to be rolled for together, rather than rolling for each unit separately. This brings the precision to the level of unit types, if not individual units. The only material difference between one fighter and another would be how far it has travelled to get to the battle, so that's all that's given up by rolling them all together.
However, if some bombers are carrying paratroopers and some are not, there is as significant a difference between them as there is between a fighter and a bomber. They must be rolled for separately.can, in my opinion, be read as the primary intent being individual firing, thus truly random casualties selection. Just in case they are all of a same type, for easy of play, I'm allowed to roll them all together, if I so prefer, but maybe I can still decide to roll them individually, not to lose the chance to kill a same type with more movement left. Can you confirm this interpretation is outright wrong, and, if I have, for example, 2 fighters, one with 3 and one with 1 movement left, I'm obliged to roll them both toghether, letting my opponent obviously taking the one with 1 movement left first, instead of rolling them individually, if I so prefer?
What does krieghund exactly mean when he says "the rules allow for"? Does he mean that is merely optional (to speed things up when it doesn't matter) or mandatory like in LHTR?
Sorry for asking you basically the exact same thing you already answered. Just wanting to be absolutely sure here. -
@cernel said in AA revised minor bug:
...if I have, for example, 2 fighters, one with 3 and one with 1 movement left, I'm obliged to roll them both toghether, letting my opponent obviously taking the one with 1 movement left first, instead of rolling them individually, if I so prefer?
What does krieghund exactly mean when he says "the rules allow for"? Does he mean that is merely optional (to speed things up when it doesn't matter) or mandatory like in LHTR?
Sorry for asking you basically the exact same thing you already answered. Just wanting to be absolutely sure here.No, v3 and v4 did not introduce any optional rule here. You roll against both attacking fighters together and the attacker chooses.
No need to apologize. Your questions help me a lot to assure to cover (hopefully) every aspect of this complex topic with its development over different editions/rulesets.
-
@panther I wish also to point out that the adoption of the true Revised (since OOB, apparently, albeit well explained only with LHTR) casualty selection, in substitution of the currently purely random one (a behavioural change due to be done, on the account that we have clarified this is the will of the creator), will add a number of additional challenges, on which you may want to give your opinion, as mostly a matter of keeping coherency. For example, in TripleA, AA gun style units may do their peculiar shots in attack too. This would add the element that you may have a same type of unit belonging to different players (world powers), and, in such a case, it will need to be decided whether a same type of units having different ownerships is to be accounted for the same group or not. My suggestion, on this account, would be that different ownerships dictate separate grouping, also since this will avoid introducing the (currently generally unsupported by TripleA) element of multi-player casualties selection agreement.
Again, I surely suggest keeping the current behaviour of truly random casualties selection as a property (I guess an additional one, since the current ones will need to be corrected), so that the mapmakers may have it, if they so wish.
On this matter, the random casualty selection for AA currently has what I would define a display bug. During the course of a battle, you have no way to know which ones of your AA hit units with different remaining movements have been shot down, that may be relevant for subsequent casualties decision (it sometimes happened to me to have this issue when I was attacking Caucasus with Germans in World War II v3 1941, having air units shot down by AA gun fire, without then fully knowing what remaining movement my surviving air units had, influencing my casualties choice (for example, not knowing if that fighter with 1 movement left that I would have to land on an exposed territory was still there in the battle or happened to be the one shot by the AA gun fire)).
-
@cernel said in AA revised minor bug:
@panther I wish also to point out that the adoption of the true Revised (since OOB, apparently, albeit well explained only with LHTR) casualty selection, in substitution of the currently purely random one (a behavioural change due to be done, on the account that we have clarified this is the will of the creator), will add a number of additional challenges, on which you may want to give your opinion, as mostly a matter of keeping coherency. For example, in TripleA, AA gun style units may do their peculiar shots in attack too. This would add the element that you may have a same type of unit belonging to different players (world powers), and, in such a case, it will need to be decided whether a same type of units having different ownerships is to be accounted for the same group or not. My suggestion, on this account, would be that different ownerships dictate separate grouping, also since this will avoid introducing the (currently generally unsupported by TripleA) element of multi-player casualties selection agreement.
Do you have an example scenario that would help me better understand this case?
Again, I surely suggest keeping the current behaviour of truly random casualties selection as a property (I guess an additional one, since the current ones will need to be corrected), so that the mapmakers may have it, if they so wish.
Do you by chance know why the casualty selection has been implemented this way? I mean there have never been rules for a random selection - so perhaps some simplification?
On this matter, the random casualty selection for AA currently has what I would define a display bug. During the course of a battle, you have no way to know which ones of your AA hit units with different remaining movements have been shot down, that may be relevant for subsequent casualties decision (it sometimes happened to me to have this issue when I was attacking Caucasus with Germans in World War II v3 1941, having air units shot down by AA gun fire, without then fully knowing what remaining movement my surviving air units had, influencing my casualties choice (for example, not knowing if that fighter with 1 movement left that I would have to land on an exposed territory was still there in the battle or happened to be the one shot by the AA gun fire)).
Yes, I have noticed that, too.
-
@cernel I just did a quick test how AA-fire casualty selection is resolved in wwII_v5 and wwII_global. While the attacker can correctly choose between different air units, the remaining movement is not displayed. However it appears that the engine automatically takes out the chosen air unit with the least remaining movement points - unless it had not been a coincidence . So in these games casualty selection appears to be resolved reasonably (unless further tests prove the opposite).
-
@panther said in AA revised minor bug:
@cernel I just did a quick test how AA-fire casualty selection is resolved in wwII_v5 and wwII_global. While the attacker can correctly choose between different air units, the remaining movement is not displayed. However it appears that the engine automatically takes out the chosen air unit with the least remaining movement points - unless it had not been a coincidence . So in these games casualty selection appears to be resolved reasonably (unless further tests prove the opposite).
You mean the defender, right? Anyways, yes, when you can choose, the air units with less remaining movement are taken first, in any case you can. This (also for consistency) is what I would suggest in the matter at hand, as well, since, while, by rules, one could decide to kill the unit with the most movement left first, that would make no sense as a choice; so I think it's opportune the engine going ahead, making the obvious choice, when only one there is.
Regarding the AA firing for the attacker, thus possibly having same type targets belonging to different players, that is something relevant, as, in custom games, you may have that, as, in TripleA, you can have AA firing against defending units too (like you can even have aircrafts making AA shots against AA guns). This is just one of several cases that will be necessary to take into account, when correcting this bug. Another matter, for example, is that, while in the basic games all AA targets have 1 hitpoint, in TripleA you can have air units with multiple hitpoints or AA targeting not-air units with more than 1 hitpoint. So, in this case, if I have, like, two dragons that are under AA fire, one with 1 damage and the other one undamaged, and I'm using v2/v3/v4 casualties selection, should they count for different groups, when they currently have different hitpoints left, or should the dragons' owner be able to decide what dragon is taking the damage, by having them part of the same group, no matter how much damaged? The other matter is that, while in the basic games you have something like v5, in which you have a limited number of AA shots (up to 3 per AA guns, instead of infinite) and full casualties selection, in a custom TripleA game you can have limited number of AA shots, like in v5, but inter-type random assignation of hits (currently fully random, but this is the bug), like in v2. That would be another challenge, about how that should actually work (If I have 6 AA shots with 13 eligible targets divided into 4 group types, who is going to get the hits, at the end?), in the moment in which the hit assignation is not anymore purely random per specific unit, like instead it (wrongly) works now. And it will be very important to fully document how everything works, in pos2. I hope the developers will make sure of that.
-
@cernel said in AA revised minor bug:
You mean the defender, right?
No, the attacker chooses to take out the air unit casualties shot down by the defending AA gun(s).
Anyways, yes, when you can choose, the air units with less remaining movement are taken first, in any case you can. This (also for consistency) is what I would suggest in the matter at hand, as well, since, while, by rules, one could decide to kill the unit with the most movement left first, that would make no sense as a choice; so I think it's opportune the engine going ahead, making the obvious choice, when only one there is.
Agreed!
Regarding the AA firing for the attacker, ...
Thank you, I understand that better now. I will think about it and add my thoughts soon.
-
@panther said in AA revised minor bug:
@cernel said in AA revised minor bug:
You mean the defender, right?
No, the attacker chooses to take out the air unit casualties shot down by the defending AA gun(s).
Right, got mixed up a bit there. Eh, I wouldn't mind an option for letting the AA firer choose casualties, but this would be an off topic feature request.
-
@cernel said in AA revised minor bug:
Regarding the AA firing for the attacker, thus possibly having same type targets belonging to different players, that is something relevant, as, in custom games, you may have that, as, in TripleA, you can have AA firing against defending units too (like you can even have aircrafts making AA shots against AA guns). This is just one of several cases that will be necessary to take into account, when correcting this bug. Another matter, for example, is that, while in the basic games all AA targets have 1 hitpoint, in TripleA you can have air units with multiple hitpoints or AA targeting not-air units with more than 1 hitpoint. So, in this case, if I have, like, two dragons that are under AA fire, one with 1 damage and the other one undamaged, and I'm using v2/v3/v4 casualties selection, should they count for different groups, when they currently have different hitpoints left, or should the dragons' owner be able to decide what dragon is taking the damage, by having them part of the same group, no matter how much damaged? The other matter is that, while in the basic games you have something like v5, in which you have a limited number of AA shots (up to 3 per AA guns, instead of infinite) and full casualties selection, in a custom TripleA game you can have limited number of AA shots, like in v5, but inter-type random assignation of hits (currently fully random, but this is the bug), like in v2. That would be another challenge, about how that should actually work (If I have 6 AA shots with 13 eligible targets divided into 4 group types, who is going to get the hits, at the end?), in the moment in which the hit assignation is not anymore purely random per specific unit, like instead it (wrongly) works now. And it will be very important to fully document how everything works, in pos2. I hope the developers will make sure of that.
Apart from the core game principles we have already discussed in this thread, your scenarios add another core game principle of A&A games: the multinational defense. In A&A games - as you know - if the attacker scores a hit, the defending players will have to agree about the casualty. If they cannot agree, the attacker chooses. In TripleA you usually see all valid units of all powers involved in the battle screen and can choose the unit you want to take out.
Concerning units with two hitpoints we have that situation in regular battles against battleships. Here the defender is (or the defenders are) totally free whether to assign two hits e.g. to two battleships or to sink one battleship. It is always the choice of the opponent(s) how to assign hits.
Regarding AA-fire (or better "AA-like-fire") I don't see a reason to negate these principles.
All in all I would recommend to resolve this as follows:
- Roll against groups of units if possible (ignoring the defending power and ignoring already assigned hitpoints)
- Among those groups let the defender(s) choose. In case there is more than one defender let all of them agree. In case of two-hit-units let the defender(s) choose how to assign hits, too.
In v5 and Global1940 games this is easier: The number of shots is determined by the number of AA-guns or targets, and (unless special rules apply) the opponent is free to choose.
IMHO this matches the A&A rules as close as possible.
-
@Cernel Unfortunately there appears to be a random factor in casualty selection in v5 (so most likely in 1940_any as well), too. In the attached savegame, 5 bombers (with remaining movement points 1,2,3,4 and 5) bombed the Moscow factory. The AA-gun shot down two of them and the engine removed a bomber with 1 movement point left and a bomber with 4 movement points left. So we have an issue here, too.
0_1538464873078_v5test_2a.tsvg -
So all in all we have a thread that started as "AA revised minor bug" and that turned into identifying and describing some major issues affecting all versions (except maybe v6).
-
@Panther or other, is there another bug report for the major issues uncovered? Could those issues be summarized for a bug report?
-
@LaFayette Yeah, this thing really needs to be fully summarized in a bug report. One of me or @Panther could do that, I suppose, and the other one double check it, but it would be quite a lot of work, for something that maybe will be just closed off in 6 months. Anyways, I'll check it out, if anyone does it.
-
I mean, if there is a developer that it is interested in looking into this bug, to fix it, I can see to summarize it, as it is really something of some importance, but only if also the current behaviour is going to be kept, as an option, even if maybe not currently used by any maps, as I think the official one is really strange.
EDIT: We could keep the current one working for pos2, as, since that map was made by veqryn, we can be pretty sure the behaviour was intended.
-
@Panther said in [Open] AA revised minor bug:
So all in all we have a thread that started as "AA revised minor bug" and that turned into identifying and describing some major issues affecting all versions (except maybe v6).
Yes. All versions but v1 and v6 are surely affected.
v6 is surely not affected.
v1 we still have to clearly define it; so this issue cannot really be fully summarized until that is covered too, actually. -
@Cernel said in [Open] AA revised minor bug:
for something that maybe will be just closed off in 6 months.
Ice box is harsh, but compare that to this bug which has not even been looked at... Which would you prefer, for a tactical decision of "we're not going to get to this" compared to "whoops, we did not even see this."
The other aspect is about maintainer efficiency. We're few and very stretched. It used to be the full time job of multiple people to keep up with the project. So that we can make progress and not just tread water with responding to issues but actually fix issues, there are some efficiency gains needed.
It is also hard to say who will and can pick up the bug. If the "cost of entry" is to spend 45 minutes reading to then figure out this is something you can't work on, that is 45 minutes that could have been spent actually fixing something else. It is the case that new dev's come along and they need projects that are easy to pick up so they can become engaged with the TripleA codebase. Having bugs to require so much background hinders that.
In my professional experience and learnings, bug tracking where you have this situation is a project anti-pattern. It leads to these long bug queues that never get solved and you spend a lot of time curating bugs not actually fixing any.
So yes, we need a summary if there are items that need to be fixed, and ideally they would be placed in the official bug tracker so that the maintainers do not have to spend time bouncing between a set of forum posts and the bug tracker
-
Well, to be honest, for me there is also the issue that I believe how it works now is more realistic, or anyways I prefer it. Of course, this shouldn't matter, as nobody asked me about that, and I'm not saying I like bugged stuff, as I realize a bugged behaviour is not reliable.
And, as I said, we still have to get around clarifying v1, as I believe the official rules and clarifications there are unclear. We might have to get back to someone official there, actually, meaning moving to another forum.
-
@Cernel Okay, if there is something that needs to be fixed, we need a bug report. A bug report has four elements (it's actually kinda formal):
- status: is it open or closed
- reproduction steps: minimized list of actions to create the problem
- problem description: typically a "notice" line that says, notice this behavior, that is the bug
- expected behavior: a description of instead of the 'notice', what should have happened.
Of course, the more terse and to the point the bug report the better, short and sweet. Forums perhaps dovetails here as identifying the actual problem and discussing what the right fix can be difficult. Typically I've seen that best handled by simply re-creating bug reports to be concise summaries of longer threads, so one closes an old bug and opens a new short one (in this way the 10 pages of text is then converted to a proper bug report that then has a better chance to be picked up and fixed). The open/close status is just for efficiency, you need to be able to answer "what are the open bugs? What should I look at?". The repro steps and problem description is so the dev can create the problem and know they are looking at the same issue. The expect fix is explicit so we know how to fix, it can be not obvious what the right fix is. Software is stupid exact, exact steps of what should be done are required. This then combines with the repro steps so that a dev can then repro with a fix to ensure the problem was actually fixed.
Those 4 items define a bug report. As a software project, that is what is needed for us to be able to effectively fix bugs.
I continue to be disturbed we have a 'bug' section in forums as none of those items are properly tracked in a conversation thread, they all need to be simulated by manipulating titles and the 'prompt' of this is the information we need can only come from a 'read this first' post. It can be made to work, but it's for sure simulating proper bug tracking software in forum (aka conversation) software.
-
@LaFayette If this whole matter would be summarized in a way to be complete and understandable (that would really require a number of examples) (which we cannot do until v1 is clarified too), it will be a really long wall of text, and still require the best part of a hour to be fully understood by someone having no idea about it.
Hello! It looks like you're interested in this conversation, but you don't have an account yet.
Getting fed up of having to scroll through the same posts each visit? When you register for an account, you'll always come back to exactly where you were before, and choose to be notified of new replies (either via email, or push notification). You'll also be able to save bookmarks and upvote posts to show your appreciation to other community members.
With your input, this post could be even better 💗
Register Login