A typical server written in Makiki calls start-http-server in the
main procedure to process requests. It won't return until the server
is terminated, so if you call it in REPL, you can't interact with
the process any further.
The makiki.dev module improve the situation. It runs the server
in a separate thread so that you can keep working in REPL, and also
it supports reloading of the script so that you can incrementally
improve the program without stopping the server.
Load the server script specified by SCRIPT-PATH and runs the
server in a separate thread.
It remembers SCRIPT-PATH, so you can omit it in the subsequent
calls. It also watches updates of the script file, and whenever
the file's mtime changes, it automatically reloads it.
The script is loaded into a module makiki.user, to avoid interfering
with whatever you're doing in the default user module.
If the script creates its own module and defines main (or dev-main;
see below) in it, you need to tell the name of the module with MODULE
argument. It is remembered as well, so you can omit it in the
subsequent calls.
Running the server is a bit tricky; typically, Makiki server script
creates and initializes an application object and then calls
start-http-server in main procedure.
We don't know how you construct the application object,
so we merely call main within a separate thread. The thread exits
when main returns, which typically means start-http-server exits.
If you need a different setup for development rather than the actual
deployment, you can define dev-main procedure in the script as well.
If dev-main exists, start-server! calls it instead of main.
If the server is already running, it does not restart the server thread. However, before loading the script, it empties http handler database, so reloading the script replaces entire handlers.
Since reloading doesn't call main or dev-main again, changes
in the application object may not be reflected by mere reloading.
In such a case, you need to call stop-server! and then start-server!
to reinitialize application and restart the server loop.
Stops the currently running server loop. If a server loop isn't running, this is no-op.
Stopping the server is done via implicit server control channel
(see make-server-control-channel and terminate-server-loop in
README.md.) So if there're pending requests, you
server loop processes them before exitting.