• Simultaneous games?

    Pinned
    6
    2 Votes
    6 Posts
    3k Views
    LaFayetteL
    I can say it is somewhat common to observe a game (or two) while waiting in a staging window or the lobby for a game of your own to start. @RoiEX it would be good to know a couple things as well specifically regarding the updated behavior: A) would the lobby window closes when connecting to a game? Currently it does not, would this change? B ) assuming the lobby window stays open, what would happen if a user did try to connect to a second game We can maybe pull stats from DB for how often a player connects to multiple games, I don't think it is rare per-say. It is a question though whether the "hassle" to fire up another TripleA instance manually, login to lobby again, all to join a second or a third game would be a major annoyance.
  • Poll on how Defender Retreat should work

    23
    2 Votes
    23 Posts
    706 Views
    I
    This is my interpretation of the rules & what I coded: At the very start of the round, if any unit has the attachment canDefensiveRetreat = true, the defender gets a prompt if they want to retreat, and to which territory. They may only retreat to friendly/neutral territories, not an enemy territory, and they can't retreat to a territory with enemy units in it, unless only enemy air is present, which is non-blocking. If the defender chooses not to retreat, then the battle proceeds as normal. If the defender chooses a territory to retreat to, then that territory is saved and stored for later. The attackers fire at all of the defending units. Any units that are doing a defensive retreat do not fire back. Casualties are removed, and then any retreating units retreat to the destination we stored earlier. Any defender units not capable of retreating remain and must fight to the end/get captured/get removed, whatever happens normally EDIT: I was originally going to have a property that applies to all land units, but it turns out that in A&A North Africa not all land units can defensive retreat (the truck and the supply token can't) so it pretty much needs to be a unit attachment. This offers some variation as the mapmaker can select what units they want to be able to do a defender retreat
  • Everything now auto-deploys on merge to main!

    2
    4 Votes
    2 Posts
    110 Views
    LaFayetteL
    Notable mention, bot hosts are an exception. They need a manual docker pull & restart to pull in changes. This is largely intentional as I would consider them already as legacy. edit: plus, restarting a live game is incredibly disruptive. I wasn't comfortable having the lobby restart without warning until there was the auto-reconnect capability implemented - let alone interrupting a full blown in-progress game.
  • why did we remove the undo all button for movement phases?

    20
    0 Votes
    20 Posts
    761 Views
    ubernautU
    @Cernel wasn’t even aware of all of these existing commands i do use the control click for getting an entire unit type. didnt realize at all that right clicking current deselects units selectively. i dont think it’s a great idea to flip the click tho not sure the current thing is broken just a bit convoluted and not well understood probably
  • Naval Models & Colors ?

    3
    1 Votes
    3 Posts
    141 Views
    B
    @jayc707 Hi JayC The images will be much sharper in game then they appear when posted here as well. The UHD maps use mostly Frostion units also , as well as his Iron War and Age of Tribes. You can add units to your preferred map and customize it. The map making threads have the information on how to do it. Ask there if you are interested in learning how. Donations are always welcome
  • Game Engine Rules (AI Training)

    22
    5 Votes
    22 Posts
    2k Views
    RogerCooperR
    @Kindwind Any further progress on this. I would like to see a report of an actual game. Even seeing the first turn would be enough to see whether you are getting somewhere. Note that OOB Axis & Allies 1942 (2nd edition) strongly favors the Axis. It would be better to use the Larry Harris alternate setup.
  • Switch to "FlatLaf" for Look & Feel (updated skin)

    15
    4
    1 Votes
    15 Posts
    818 Views
    PantherP
    I like Darcula - except the politics panel - those colours are a pain, indeed.
  • 4 Votes
    26 Posts
    1k Views
    LaFayetteL
    I suppose it is worth pointing out that a 'revert' in git is just an undo update. The code history in Git is write-once. Git never deletes anything. So, in a revert update, based on the reverted change: any lines that were added are removed, and any removed lines are re-added. So, a revert is a new update, it just an inverted update to cancel out a previous one. Revert PRs are not at all special to Git, it's just another update. Once a revert is merged, a person can revert the revert update, add more changes to it, and then merge all of that as one bundle. So, a flow might be like this (the letters on the left hand side represent the commit ID, AKA the commit SHA): a43ae Some update that introduced a bug 3ads2 Revert: Some update that introduced a bug fae54 Revert: Revert: Some update that introduced a bug + Bug Fix Changes to the code base are "squashed" together. When a person clicks the revert for the revert, that creates a new proposed change. Bug fixes can be added to that proposed change, then it'll all be squashed together as a single commit that is then released. For example, when reverting the revert, this change would be proposed: abc12 Revert: Revert: Some update that introduced a bug In that proposed change, we can add the bug fix, eg: abc12 Revert: Revert: Some update that introduced a bug fea23 Bug fix Then, when merged, that would land on the main branch and in the history as one commit, eg: fae54 Revert: Revert: Some update that introduced a bug + Bug Fix
  • Urgent communication needed - 2026-06 Launch & MARTI Dice Server

    Moved
    41
    2 Votes
    41 Posts
    1k Views
    PantherP
    @Nothingtoseehere You will need to use workarounds here or play at https://www.axisandallies.org/forums/ PBF works fine there, as the forum-software is on a lower (but compatible) version than here.
  • Regression: Game-ID missing in new Dice Server

    4
    1 Votes
    4 Posts
    183 Views
    PantherP
    Everybody feel free to join the discussion here: https://github.com/triplea-game/triplea/discussions/14596
  • Compatibility with MacOS Silicon

    19
    4 Votes
    19 Posts
    751 Views
    ubernautU
    @LaFayette ok so maybe all that wud be needed at this point then since the menu is icons i’d reverse the up arrow that wud make it more obvious that thats where you go to download
  • Generic anthem anyone?

    8
    2 Votes
    8 Posts
    398 Views
    B
    @VictoryFirst How bout "Onward Christian Soldiers" https://www.hymnal.net/en/hymn/h/871 Edit Or some sort of drum beat https://audiojungle.net/search/war drums but yea, the air raid kinda harsh. Not to mention it's still used to this day. Don't wanna freak anyone out in a current warzone
  • is 2.5.22294 the latest release?

    7
    2 Votes
    7 Posts
    559 Views
    P
    @thedog downloaded the 393 and started a game. It's working great and the AI is much faster. Now I have a more challenging game and stronger allies & axis bots.
  • Is TripleA finally dead

    22
    2 Votes
    22 Posts
    3k Views
    LaFayetteL
    @johnnycat you raise good questions: Is the answer time and programmers? Or just programmers or just time? Or management ? Does it come down to Money? If the answers are: yes, no, maybe, and only on weekends, how could I and others help? Time and programmers is the biggest need. Every bug fix and issue probably takes 2 to 16 hours to fix. This is the biggest gap. Further, TripleA is a larger code base that is very brittle. Making that codebase be easier to work with would help. Money is useful for keeping the servers running and paying the monthly bills. As much as I would love to be payed for TripleA work, paying any type of reasonable developer wage becomes way too much very quickly. Additional management would be helpful. Getting issues sorted, cleanly stated, boiled down to a bare minimum would all be helpful.
  • ML and Cloud deployment

    4
    1 Votes
    4 Posts
    583 Views
    RogerCooperR
    @kindwind It would seem that training against random games would be every inefficient compared to training against the AI that TripleA comes with.TripleA is not Go, random play should be very bad.
  • Looking for beginner-friendly issues to start contributing

    5
    3 Votes
    5 Posts
    784 Views
    TheDogT
    @Darmok72 here is a new starter request, it was half done but not completed, it will affect about 1/4 maps, so please attempt this. https://forums.triplea-game.org/topic/2999/2-6-release-getting-close-need-volunteers-to-help-beta-test-2-6/73?_=1762176127175 EDIT: Also copied links from above link https://forums.triplea-game.org/topic/2752/maxbuiltperplayer-error-resets-purchase-ui-was-how-to-limit-the-number-of-units-purchased/11 https://github.com/triplea-game/triplea/issues/9231
  • Does anyone have a praser for Triplea?

    2
    0 Votes
    2 Posts
    2k Views
    RogerCooperR
    @kindwind Efforts You seem to have in mind using a neural net style AI to play TripleA. I suspect that this we actually be quite difficult given the huge number of possible board positions.You might be better off using a very small mod like Minimap even without a large record of games to train it. TripleA's own AI can be used for training.
  • Dev Status Update (2.7)

    5
    3 Votes
    5 Posts
    1k Views
    LaFayetteL
    @ubernaut yeah, redoing MARTI would potentially save us $10/month on costs as well. I think redoing the bots with 'network relay' might be higher up on the list. That will save a ton of code, create an effective infinite number of bots, and save us maybe like $25/month. The 'network relay' will be an update where the bots will just be responsible for only forwarding network packets. Today, a bot is basically an entire player but is made somewhat invisible. We are bounded by the amount of RAM on a host for how many bots we can host. If bots are just sending network packets, they'll take up very little RAM. We'll only be bounded by network traffic. We could theoretically run dozens of bots per host instead just 3 to 5. This should cut our bot host count down from the current 6 or 8 today to maybe 2. I think there might be some quick wins with the server's as well. AT least for me, I might be inclined to do a quick ELO system before redoing MARTI.
  • New Lobby Release Model for 2.7+ / Zero-Downtime & Continuous Release

    1
    2 Votes
    1 Posts
    582 Views
    No one has replied
  • 1 Votes
    10 Posts
    2k Views
    LaFayetteL
    @rogercooper Not yet. I'm thinking to bump the version number to '2.8' when all of these updates are ready. The '2.8' release announcement will be when it is ready. These backend changes are most of what we are waiting for.

Recent Posts

  • This is my interpretation of the rules & what I coded:

    At the very start of the round, if any unit has the attachment canDefensiveRetreat = true, the defender gets a prompt if they want to retreat, and to which territory. They may only retreat to friendly/neutral territories, not an enemy territory, and they can't retreat to a territory with enemy units in it, unless only enemy air is present, which is non-blocking.

    If the defender chooses not to retreat, then the battle proceeds as normal. If the defender chooses a territory to retreat to, then that territory is saved and stored for later. The attackers fire at all of the defending units. Any units that are doing a defensive retreat do not fire back. Casualties are removed, and then any retreating units retreat to the destination we stored earlier. Any defender units not capable of retreating remain and must fight to the end/get captured/get removed, whatever happens normally

    EDIT: I was originally going to have a property that applies to all land units, but it turns out that in A&A North Africa not all land units can defensive retreat (the truck and the supply token can't) so it pretty much needs to be a unit attachment. This offers some variation as the mapmaker can select what units they want to be able to do a defender retreat

    read more

  • read more

  • @iratoric which North Africa game? AFAIK some of the variations are mutually exclusive. I think the first task it is to identify which variations should be plausibly supported. Once done, I would like a full spec identified. Following, we can then look at the plan for how that would be encoded into XML. Doing so iteratively and not all-at-once is perfectly fine. Though, I think we should know beforehand which retreat variations we are likely to want (and also in which maps).

    read more

  • @iratoric said:

    Thanks everyone for your opinions. I have thought about this a lot, and it is probably best if we start with implementing one set of rules and branch out from there. Otherwise there are simply too many possible combinations to do all at once. Since the North Africa game is the “official” ruleset and an already existing game that needs this rule, I think that whatever I implement should support that first. I will code it in such a way that it still leaves room for more variations later. Thoughts?

    I instead think that the baseline should be some version of what you/we can devise as the simplest retreat system, have that done and tested, then adding properties and/or options for each element adding complexity.

    For example, in the basic system you retreat to the same zone. Next step, with property, is retreating to an other zone. Next step, with property, is limiting retreat only to zones not under attack by land units. And so on.

    read more