Go to file
Omar Polo 72342dc960
implementing CGI – NOT READY YET!
This is a first try at implementing CGI scripting.  The idea is that,
if CGI is explicitly enabled by the user, when a user requires an
executable file instead of serving it to the client, that file will be
executed and its output fed to the client.

There are various pieces that are still lacking, the firsts that comes
to mind are:

 - performance: the handle_cgi just loops ignoring the
   WANT_POLLIN/POLLOUT and blocking if the child process hasn’t
   outputted anything.
 - we don’t parse query variable (yet)
 - we need to set more variables in the child environment
   side question: it’s better to set the variables using setenv() or
   by providing an explicit environment?
 - document what environment the CGI script will get
 - improve the horrible unveil/pledge(cgi ? …)

but now I can serve “hello-world”-tier script from gmid!
2020-11-06 13:01:31 +01:00
.gitignore initial commit 2020-10-02 19:39:00 +02:00
ChangeLog added option to log to a file 2020-11-06 10:58:30 +01:00
gmid.1 implementing CGI – NOT READY YET! 2020-11-06 13:01:31 +01:00
gmid.c implementing CGI – NOT READY YET! 2020-11-06 13:01:31 +01:00
INSTALL.gmi extend installation notes 2020-10-03 14:13:55 +02:00
LICENSE added license 2020-10-02 19:54:59 +02:00
Makefile improve make error message if etags is not found 2020-10-07 17:59:55 +02:00
README.md implementing CGI – NOT READY YET! 2020-11-06 13:01:31 +01:00

NAME

gmid - dead simple zero configuration gemini server

SYNOPSIS

gmid

**-hx**]
\[**-c** *cert.pem*]
\[**-d** *docs*]
\[**-k** *key.pem*]
\[**-l** *access.log*]

# DESCRIPTION

**gmid**
is a very simple and minimal gemini server.
It only supports serving static content, and strive to be as simple as
possible.

**gmid**
will strip any sequence of
*../*
or trailing
*..*
in the requests made by clients, so it's impossible to serve content
outside the
*docs*
directory by mistake, and will also refuse to follow symlink.
Furthermore, on
OpenBSD,
pledge(2)
and
unveil(2)
are used to ensure that
**gmid**
dosen't do anything else than read files from the given directory and
accept network connections.

It should be noted that
**gmid**
is very simple in its implementation, and so it may not be appropriate
for serving site with lots of users.
After all, the code is single threaded and use a single process
(multiple requests are handled concurrently thanks to async I/O.)

If a user request path is a directory,
**gmid**
will try to serve a
*index.gmi*
file inside that directory.
If not found, it will return an error 51 (not found) to the user.

The options are as follows:

**-c** *cert.pem*

> The certificate to use, by default is
> *cert.pem*.

**-d** *docs*

> The root directory to serve.
> **gmid**
> won't serve any file that is outside that directory.

**-h**

> Print the usage and exit.

**-k** *key.pem*

> The key for the certificate, by default is
> *key.pem*.

**-l** *access.log*

> log to the given file instead of the standard error.

**-x**

> Enable CGI scripts.

# CGI

If CGI scripts are enabled, when a file requested by a client is
marked as executable it is executed and its output fed to the client.

Note that since this give the chance to anybody to execute possibly
**any file**
in the served directory, this option is disabled by default.

# EXAMPLES

To quickly getting started

	$ # generate a cert and a key
	$ openssl req -x509 -newkey rsa:4096 -keyout key.pem \
	        -out cert.pem -days 365 -nodes
	$ mkdir docs
	$ cat <<EOF > docs/index.gmi
	# Hello world
	test paragraph...
	EOF
	$ gmid -c cert.pem -k key.pem -d docs

now you can visit gemini://localhost/ with your preferred gemini client.

# CAVEATS

*	it doesn't support virtual hosts: the host part of the request URL is
	completely ignored.

*	it doesn't fork in the background or anything like that.