Dunedan

I have to admit that the Terms of Use aren't very clear regarding this, but I would argue point 8 and maybe point 10 as well apply. Such bots might also introduce additional security and privacy risks for our users and that's why we generally don't allow them. To be clear: Those are not artificial rules we made up for your bot, but something we have applied in the past as well, whenever somebody added a third-party bot to the multiplayer lobby. One difference of your bot, compared to other bots I have seen in the past, is that it implements functionality I can very well see to be integrated into 0ad itself (albeit I'd prefer a different interface between players and the dedicated server). The correct process is to go through the regular PR review process for that though and get it accepted into 0ad.

Please mind that we consider any player operated bots in the multiplayer lobby to be a violation of the Terms of Use.

In r28 the font engine got replaced, so you could use any FreeType or OpenType font. All you have to do is add it in the right location (it's binaries/data/mods/mod/fonts/ by default, but should work with any mod too) and change you config to use them. Here is how that'd need to look in your config: https://gitea.wildfiregames.com/0ad/0ad/src/commit/6ca52a0b23a311910e259cddf4ddab747374f22d/binaries/data/config/default.cfg#L659-L710 If you are stuck with Hyrule Conquest on a27, check out source/tools/fontbuilder2/fontbuilder.py for how to build the font files.

The problem is that without retrieving reports from all players, we don't know if the game is still in progress. As there is no artificial limit to the duration of games, we'd have to wait a few hours to be reasonably sure it's over.

To rate games we currently rely on both players "agreeing" about the outcome of the game. That means whenever a rated game ends, 0ad sends a report about its outcome to a bot which is responsible for handling the ratings. Only if the reports of both players match, the game will get rated. This is necessary to prevent players from sending fake reports to manipulate their ratings. In case one of the players disconnects and doesn't come back later, we won't get a report from that player and therefore can't rate the game automatically. The best solution I've come up with so far is to inform the bot already when a rated game starts and about which players participate. Then if only a single report gets received later on, we can wait a few hours and award that player the points. I've described that in more detail in https://gitea.wildfiregames.com/0ad/0ad/issues/3951#issuecomment-62016 The non-host player loosing the connection and not being able to reconnect afterwards (no matter for what reason) is an interesting edge case I haven't thought about yet.

As soon as a release candidate gets published, mod authors should take care to update their mods. From this point on breaking changes are unlikely and a release candidate is a clear sign that a release is about to come out soon.

The lobby isn't resource heavy and right now we should be able to handle a few thousand concurrent players. There is also work in progress to make it even more scalable (https://gitea.wildfiregames.com/0ad/0ad/pulls/8679). However, as @Stan` already mentioned, with a sudden significant increase of players we'd definitely get trouble moderating the lobby with the capabilities we have at hand for that right now.

Good catch @yeager! I created a PR to fix that, however we'll have to see if it'll make it into r28: https://gitea.wildfiregames.com/0ad/0ad/pulls/8668

Shouldn't the sorting also depend on the selected formation?

Seems like source/tools/i18n/update_templates.py contains a bug which includes leading and trailing line breaks and tabs when extracting strings from XML files. While a fix should be straight-forward, as we're in string freeze we want to pull the fix into main only and have a separate r28-only workaround for including line breaks and tabs when looking for translated strings from XML files.

Seems like it's broken for all languages, so something going wrong during the extraction of the translations seems likely.

What all of you seem to forget is that the instructions to compile a23 are quite different from the instructions to compile recent versions of 0ad and that the versions of some libraries a23 depends on aren't available on modern Linux distributions. In my opinion that makes compiling a23 on a modern Linux distribution quite complicated.

There have been some changes to the lobby for r28, which could receive wider testing before we release, so it'd be great if some of you could test multiplayer games using the lobby with the release candidate. What I'm interested in are answers to the following questions: Does connecting to the lobby work as usual? (verify that lobby.tls and lobby.verify_certificate aren't set in your user.cfg or that they're both set to true (which is the default for r28) Are there any changes in your ability to host games, compared to a27.1?

While translating I noticed a historical inaccuracy. Somewhere r28 contains the following sentence as part of a string: According to Wikipedia that battle happened 338 BC: https://en.wikipedia.org/wiki/Battle_of_Chaeronea_(338_BC)

The upcoming release 28 includes a complete overhaul of the text rendering. You might want to try out its current release candidate to check if that further improves the situation (don't forget to disable the mod before doing so).