In V1 I could use http.cgi to run a simple bash/perl/python script and serve the output without having to deal with a reverse proxy and making sure the other server is always running.
Is this still possible in V2, or do I now need a separate cgi server to be running at all times, hooked up via a reverse proxy? I feel like this adds an additional “moving part” that shouldn’t be necessary for really simple stuff like having an endpoint to print some kind of stats or serve a random file from a directory
Unfortunately, circumstances prevent me from porting the CGI plugin to version 2 and then maintaining it. However, this is a very simple plugin as the Go standard library does all of the real work. As I remember, the bulk of the code is spent parsing the configuration and I believe this has been greatly simplified for version 2. This could be announced by the Caddy team as a good first-time contributor project.