jonbaer

If I go back on the builds I eventually end up running into what looks like a Spidermonkey related issue, even if I trash the library and rebuild I end up getting this ... Undefined symbols for architecture x86_64: "void ScriptInterface::ToJSVal<long>(JSContext*, JS::MutableHandle<JS::Value>, long const&)", referenced from: bool ScriptInterface::SetProperty<long>(JS::Handle<JS::Value>, char const*, long const&, bool, bool) in libengine.a(Replay.o) ld: symbol(s) not found for architecture x86_64 Usually what I end up doing is git pull <revision> then ./clean-workspaces, ./build-workspaces, make clean, make build. I have already tried multiple CUDA drivers, so I am stumped. I would seriously not be surprised if this was something due to an Xcode related issue of some type. I have gone back to playing the release build instead since it's a bit hard to play w/o territory markers.

I don't think there are any nightly builds for OSX (I could be wrong) ... just to confirm I threw away my SVN copy and grabbed a GIT copy, built OSX libraries, updated workspaces, make -j3 and still no territory markings, then tried the Official Release DMG and there were marking (so it wasn't the CUDA update like I suspected). Would be great if someone w/ a Mac could build + confirm this as an issue before digging more.

Yeah the territory boundary lines are not drawing on the terrain for some reason. I even enabled m_DebugOverlay = new TerritoryOverlay(*this); and I could see the are being computed, just the borders are not being drawn :-\ Feels like something is not being done on Init() for some reason. I would have a hard time thinking this is only an OSX issue though, does any SVN build on Windows/Linux who could verify?

I went back quite a few revisions (originally I thought this was an issue w/ a CUDA upgrade I made) but they were missing even past -r 19500 so it would have to be an OSX issue if no one else has seen this yet. They did show up in -r 19000 so I will eventually isolate it :-)

Still no luck, was going to try to trace back on these commits (would there be any other class responsible for drawing the territory?) ... https://github.com/0ad/0ad/search?o=desc&q=Territory&s=committer-date&type=Commits&utf8=✓

I'm on OSX ... trying to look through the diffs but don't see anything that would have removed the markings. Should I just to just make clean + rebuild?

SVN Revision: 20204 The territory markings seem to be missing (see attached). Unless this was moved to an option/setting(?)

Has anyone else followed what DeepMind published w/ regards to Starcraft2 AI (more specifically and if I followed it correctly) - what they did was added a listener to a running game in which case they fire back game state over Protocol Buffers and let either C++ or Python scripts control them. https://github.com/deepmind/pysc2 https://github.com/Blizzard/s2client-api https://github.com/Blizzard/s2client-proto https://github.com/davechurchill/commandcenter (This would be almost a Petra equivalent using the C++ client) What I have been studying recently are the build orders (strategies under which the commandcenter runs) because Blizzard also released a ton of replay files, assuming in which case you could parse and analyze. There are too many maps to study but for a game w/o 3 races + 2 resources I find the entire thing to be (maybe) a less of a challenge than if 0AD had the same AI running backend. Hooking into Petra's Headquarter in a similar fashion - would that be possible now w/ just JS hooks or does the way Blizzard did it w/ protobuf over the wire make more sense. I'd love to see 0AD as an AI platform but I can't tell if it's already there or something would need to added. Thoughts? (I know there were OpenAI / Gym / Universe discussions before but this seems more inline w/ how to get output/input into the game)

I like how you *need* to have an XBox Live account to beta test the PC game AND sign a full NDA :-\ https://www.ageofempires.com/age-empires-insiders-tos/ 0 A.D. and open source for life! ;-)

+1 for more AI diplomacy. To me this has been the most enjoyable thing to watch. Playing a random map w/ 4 players and myself all set to 1 team and after a while having the alliances and enemies formed. I wish there were more ways to develop/program this outside of the larger game, so far from what I gather the AI asks for the alliance too late - it's unsure what the game state is to make a judgement for when this happens. I do predict a day will come though when the AI will be able to completely auto balance everything from alliances/military/trade/spies/navy - then we are doomed :-)

I'd actually like to put this onto a GitHub pull request along w/ another patch (Random AI Settings), would that be easier? The thing is I don't see any PR(s) on the current master that is up there. I had to delete my SVN copy for needed space for work related stuff - I'd like to see the entire project move to GItHub but I do recall there being issues w/ the larger files. (Seeing a split between engine code + public mod would be great too but I know how difficult that would be to keep in sync). My other option was to just create mod(s) for both against Alpha22 + just post them to GitHub repo. Either way great job handling this project in an open source setting - I'd imagine it is not an easy task.

I think this is it ... https://trac.wildfiregames.com/ticket/4284 ... what I was more interested in at the time besides a Restart option was also a Reseed ability to regenerate a random map (in case I didn't like the layout that was generated). I may have missed it since I didn't understand what the last comment on the ticket was referring to.

Man just downloaded Alpha 22 and my Restart/Reseed option didn't show up - did my patch get rejected?

Hi, I am curious to know has anyone built a world map mod somewhere? Something akin to the Red Sea map but more global or covering a larger area. I realize it doesn't have to be to scale but someone was looking for a gaming engine to do something like a global world map for a strategy game and thought he could use pyrogenesis pretty easily if that was the case, just wanted to see if that mod existed yet though. - Jon

The interesting metric I saw was over a "DBD" for anyone interested ...

I found some interesting metrics in regards to build order optimization over the weekend, question in pertaining to 0ad (as most of the research papers reflect StarCraft to a large extent) is how to view a more common build order benchmark within the game, for example most SC/SC2 orders are grouped to strategies and you can find them all over labelled as such (one example): 7 Roach Rush / Expand / Lair6 Drone until 1010 Double Extractor Trick12 Overlord12 Spawning Pool11 Drone x516 Extractor (Put 3 Drones inside immediately upon completion)15 Drone16 Queen (Inject Hatchery as soon as she pops out)18 Overlord18 Drone x220 Roach Warren19 Overlord19 Roach x733 Drone34 Overlord34 Hatchery (Expand)33 Drone x336 Lair36 Extractor Can something similar be extracted from commands.txt or are there current tools outside of the game which do this? Or are there plans to maybe adding these metric types to the new charts section? Does anyone else here study this type of field/research?

While researching IDA* I came across this small project ... https://qiao.github.io/PathFinding.js/visual/

I was originally thinking of this game mode as a challenge to some reinforcement learning algorithms and to examine what the build order was w/ a static map (in other words the layout and the resource locations were always the same). I came across another project similar to what I wanted to create ... https://github.com/santiontanon/microrts ... it is the bare minimum for figuring such a task but of course they based it on the Starcraft conception of a smaller resource pool (there are only 2 in Starcraft right?) ... I think if the map changes and maps are random there can never really be a benchmark. Also (and I think many can agree) is that 300 isn't always 300 strong. So I think looking at not just time but time and strength would be good measurement.

What I would like is to bind a hot-key to a fullscreen version of the mini map - mainly for research. Seems like it should be possible to do w/ this patch no?

What does this mean if it says Average Bonus Range @ -6 meters?

I am going to take some time and answer some of your questions for sure but in the meantime I think some best/worst are also based on which maps you are playing, for example I find the Iberians fire ship to be highly effective in destroying ports easily. What I am trying to figure out when doing random civs are what the best defensive strategies are vs. siege in each civ. That is a question I would ask a pro player. Keep up the good videos they are very informative and helpful.

For those following this thread ...

I guess what I am interested in is if you can reach it faster w/ one particular civ over another and what that ratio % is. I feel like I am getting faster at it but have not timed it so that would be nice to see. I wish something like that was a build to be a bit more portable (0ad engine + 1 random map + 1 civ) @ smaller download (for research). Post back here any progress.

This would be an interesting game mode, the fastest to 300, it seems like the path to getting there gives a good benchmark for AI and ML learning. Any thoughts?

Well I meant something outside of / besides the chat message ...