Ideally, we should be able to run the server component of this project without having to involve the www/website component at all. I'm having a hard time remembering how this all works, but I think the website is supposed to set up a config for servers that it creates?
In any case, even if we read from a config file, it shouldn't be necessary to have one in order to get a new server up and running. Command-line arguments should be available to set listening address and port. Server should know where the client directory is either by itself, or by being told by the build script. You should be able to clone, run ./runserver, and have a working server. Changing the listening port / address should be ./runserver 159.65.77.62:80 or similar
Instead, I found I had to create the following config to run the server in production:
client/server_config.default.json:
{
"Host": "159.65.77.62",
"Port": 79,
"client_assets": "/root/src/buildblast/client/"
}
This is gross because:
- server_config is in client/ despite having nothing to do with client
.default appears completely irrelevantPort is set to 79 in order to get us to listen on port 80(??????)client_assets has to be manually configured (and also why is that key snake_case?)
A somewhat related issue, but it would be nice to have the website stuff in a separate folder entirely from the rest of the code. Both to provide and encourage a healthy separation from the regular server.
Ideally, we should be able to run the server component of this project without having to involve the www/website component at all. I'm having a hard time remembering how this all works, but I think the website is supposed to set up a config for servers that it creates?
In any case, even if we read from a config file, it shouldn't be necessary to have one in order to get a new server up and running. Command-line arguments should be available to set listening address and port. Server should know where the client directory is either by itself, or by being told by the build script. You should be able to clone, run
./runserver, and have a working server. Changing the listening port / address should be./runserver 159.65.77.62:80or similarInstead, I found I had to create the following config to run the server in production:
client/server_config.default.json:This is gross because:
.defaultappears completely irrelevantPortis set to79in order to get us to listen on port 80(??????)client_assetshas to be manually configured (and also why is that keysnake_case?)A somewhat related issue, but it would be nice to have the website stuff in a separate folder entirely from the rest of the code. Both to provide and encourage a healthy separation from the regular server.