Issue/#824 support for GameOptions in matchmaker queues

[Askaholic]

Loads additional params from the database and parses out game options to send in the game_launch message. My goal is for the params column to closely resemble the gameInfo table in the lua code. So something like this:

{
    "GameOptions": {
        "FogOfWar": "explored",
        "UnitCap": 1000,
        "Share": "FullShare",
    },
    "GameMods": {}
}

But it could always be expanded in the future.

Requires FAForever/db#285
Closes #824

[BlackYps]

I'm having a hard time figuring out how the server actually sends the game launch options to the client. i think mainly because my IDE can't find the declaration of this call:

server/server/ladder_service.py

Lines 456 to 458 in 2b84b87

	await host.lobby_connection.launch_game(
	game, is_host=True, options=game_options(host)
	)

[Askaholic]

Your reliance on IDE’s has been your downfall! Just grep for def launch_game and you’ll find it:

server/server/lobbyconnection.py

Lines 1030 to 1046 in 152242a

	cmd = {
	"command": "game_launch",
	"args": ["/numgames", self.player.game_count[game.rating_type]],
	"uid": game.id,
	"mod": game.game_mode,
	# Following parameters may not be used by the client yet. They are
	# needed for setting up auto-lobby style matches such as ladder, gw,
	# and team machmaking where the server decides what these game
	# options are. Currently, options for ladder are hardcoded into the
	# client.
	"name": game.name,
	"init_mode": game.init_mode.value,
	"rating_type": game.rating_type,
	**options._asdict()
	}
	await self.send({k: v for k, v in cmd.items() if v is not None})

[codecov]

Codecov Report

Merging #825 (bf9c5e7) into develop (cf4e9b0) will increase coverage by 0.04%.
The diff coverage is 95.83%.

Impacted Files	Coverage Δ
server/db/models.py	100.00% <ø> (ø)
server/ladder_service.py	97.95% <94.73%> (-0.34%)	⬇️
server/matchmaker/matchmaker_queue.py	92.02% <100.00%> (+0.17%)	⬆️
server/types.py	100.00% <100.00%> (ø)
server/gameconnection.py	95.17% <0.00%> (+0.68%)	⬆️
server/protocol/gpgnet.py	95.00% <0.00%> (+5.00%)	⬆️

[BlackYps]

This integration test fails sometimes on my machine. I saw in the log that it took many matchmaker iterations until the logged in players finally were queued up for searching. They basically matched immediatly, but that was in the teardown log already, so I believe it was after the timeout aborted the test. Maybe increasing the timeout limit would help?

Suggested change

	read_until_command(proto, "match_found", timeout=30)
	read_until_command(proto, "match_found", timeout=60)

On the other hand I couldn't get it to reliably fail with a 10s timeout either, so I don't really know what exactly is happening. Other Integration tests also fail sometimes, but that is unrelated to this PR. It seems like it sometimes takes a lot of time for a command to take effect, for no apparent reason.

[Askaholic]

Yea I think it has to do with the fast forwarding somehow. Is it really timing out here though? This should only start waiting once all of the players have successfully queued, because queue_player_for_matchmaking waits for the search_info start message before returning.

[Askaholic]

Well, I haven't gotten these to fail, so I'm just going to leave it as is for now.

[BlackYps]

I'll just keep an eye out on this. Maybe it has to do with the local configuration on my machine...