- Setup:
- Installs for the most part went fine, which is a minor miracle. The wireless network from across the house wasn't stable enough for a couple installs, but that's not the installer's fault.
- Initial content downloads from the server were slower than desirable but acceptable when on a decent connection. It's about 30 to 40 MB to connect right now; a lot of that content isn't used and could be pruned from the database. If there was a live/internal separation here it'd work better; the live version could prune itself and shrink the download. For now I'll just deal with it; the download isn't long on a LAN and I don't want to work on pruned data, or have to prune it when deploying. There's a good automatic long term solution in build scripting, but it's not worth the time right now.
- Server Side:
- The multi-zone-server seems to be working quite well. Most of the scenarios played were not launched by me manually, and were successfully executed by the universe server. I disabled the scenario-destructor so they stuck around forever, but that's fine for now too. There were a couple notable server faults:
- Sometimes units failed to save or remember their levels. This was a desync bug between the skirmish and the universe caused by thread concurrency problems. Some changes to how the two communicate led to said communication being on many threads, and someone forgot to lock important data structures. I hadn't tested multiple joins, nor players creating units while other units were live. Problem solved for now.
- Slowdowns when having more people in a game; I'll get to this when going through the scenarios played.
- In general though, the server was far far more stable than before, and required very little attention. I restarted it once to try and figure out the slowdown but that's it.
- Scenarios:
- TerraTest, the single player scenario, was largely unchanged and so worked alright. A couple people played it intermittently while waiting for others and it worked :)
- KoboldKing, the main event for the LAN, had some ups and downs
- The scenario itself worked fine, the mechanics of the enemy defense formation and special actions all worked correctly. I was hoping to see people try different tactics to break the enemy commander, and people did, even new players, which was good. People figured out basic hit and fade tactics fine and learned the basics.
- With 2 people in play, the scenario was balanced reasonably well. They won after some time (probably a bit too long, but that's easier to adjust), were patient through the too-long phase 2, then had fireworks. Yay.
- With 4 people in play, the scenario was overbalanced in the enemy favor. The number of ranged units scaled up too fast, leading to almost immediate death of any melee unit approaching range. At best a maneuver would complete, then melee units would drop. While they made progress with their hit/fades and coordinated 4 players, progress was too slow, and inevitable some players were wiped out and the scenario failed. This is a matter for tuning mostly.
This is the enemy arrangement with 4 players. The main problem here is that when a unit engages the shields on the edge, they also enter range of about 8 scouts and 8 archers. This makes melee sad.
- With 4 people in the scenario bogged down pretty bad. I think this was mostly due to a configuration bug; the multizone server ran the debug version from the release version universe server. The difference in the server case is massive so that may be all that's needed to fix. I've done some profiling since then and found a few mechanics are extremely slow with high unit counts, so it may be time to adjust said mechanics for speed. I noted during the LAN that the slowdown was in the universe server rather than the multizone server; I can't explain that quite yet, but seems important.
- I believe with some more knowledge about mechanics the players would still be able to win (or some unit levels, some of which weren't saving correctly).
- Maul, the area capture pvp scenario:
- Mechanically worked, except for a fault with terrain modification causing confusion.
- PvP is curious right now; it looks like defenses are outweighing offenses pretty strongly. It looked to me like fights moved very slowly. This could also be novice players not understanding how to utilize offense through maneuvers. Also, they could've been rallying in combat; Rally is supposed to work in combat but have reduced effectiveness, but may be currently bugged and too strong.
- Some PvP objectives would need changing; cooldowns on area capture for one. Players were spamming the claim action. As hilarious as it was to watch it's kind of a stupid mechanic to allow.
- There are camera issues around hills that aren't resolved. Clicking the ground fails if there's ground behind the camera. Needs fixin'.
- Open Field (practice area)
- The target dummies moved. That's just silly. I think some combination of historical changes in speed ended up with the dummies having positive speed. Also, I forgot to not give them an AI in the scenario editor.
- Overall:
- The overall impression that I heard was that it was more of a 'game' now than an interesting quirky demo thing. I get that feeling myself too recently, mostly since the maneuver mechanics made it in and are working well. Was good to hear. Very good.
- There are still control issues, but far far less than before. Some may be confusion about mechanics; a tutorial would be nice to see if players continue having control problems after understanding movement mechanics. Most of the frustration I heard was along the lines of 'Why did they do that', or was caused by the overbalanced commander scenario.
- So now what?
- Server speed, stability fixes, usability bug fixing, tutorial/training. More fireworks. Anything else?
No comments:
Post a Comment