Page 1 of 1

How to start the game engine without a front end

PostPosted: Thursday, 18th February 2021, 05:32
by wangster
I would like to be able to interact with crawl using the raw json. According to an older blog post (link below) the crawl engine is stand-alone and communicates with tiles and console using sockets however I can not figure out how to launch just the crawl engine.

I installed crawl-tiles from the apt-get debian package and the only binary appears to be /usr/games/crawl-tiles which only works from the desktop and launches the entire thing.

https://crawl.develz.org/wordpress/webt ... chitecture

Re: How to start the game engine without a front end

PostPosted: Saturday, 20th February 2021, 22:43
by advil
This page is describing the architecture of webtiles in particular, which isn't part of offline distributions -- regular tiles/console don't use sockets at all. To get a build that does this, you will need to build it yourself. In the repository source directory, "make WEBTILES=y" will do it, though this gives a dev-oriented setup without other configuration. See also https://github.com/crawl/crawl/blob/mas ... /README.md for some intro documentation.

There is no current version that completely lacks a front-end; the webtiles build uses a console front-end that can be controlled via a json over a socket connection (and then the console output is what online servers use for screen-recording). There is actually a branch that is not merged that fully suppresses the front-end (https://github.com/crawl/crawl/tree/headless/) but this will probably need some more work before it can be merged, not sure of the ETA. (But for developing a bot or the like, it might actually be easier to be able to see the console front-end in a terminal.)

Re: How to start the game engine without a front end

PostPosted: Sunday, 21st February 2021, 01:15
by wangster
I finally clued into the fact that TILES and WEBTILES are not the same thing and am now compiling WEBTILES.

The reason I asked the question is I'm curious about the possibility of attaching different front-end UIs using crawl as the game engine. I was hoping the main crawl game engine simply opened a json socket that the tiles/webtiles front-end connected to. That doesn't appear to be the case in the main branch but your tip about the Headless branch sounds like what I'm after. Thanks for that.