Simple HTTP is library to build web server and site. It started as a fork of
[] to experiment with OCaml
5 domain and effect to replace threads.
It provides simple routing, based on URL path, address, port or even the Host
field of the HTTP header. It can serve static files from your disk (and update
on your disk are reflected immediatly), but can also compile a directory using
a provided tool vfs_pack. Compression with camlzip is supported in both
cases (only deflate, but we plane to add gzip). Through vfs_pack you can use
.chaml files which is similar to .php, but compiled with OCaml.
It also supports server-sent events (w3c)
SSL is available and we use OCaml 5.0.0, epoll, evenfd and TCP_CORK options to get the best performance. Unfortunatly, it is Linux only, but we plan to use kqueue to support BSD, and maybe windows later.
For small request it can handle more then 1K simultaneous connection and serve more than 100K request per second. Some graph showing the performance is available on the documentation.
Out goal is to have a library able to cover the maximum use case, but simple
to use. Like Tiny_httpd, it is
still free from all forms of ppx, async monads, etc. 🙃 and the dependencies
remain relatively minimal (although there are now growing).
Note: it can be useful to add the jemalloc opam package for long running
server, as it does a good job at controlling memory usage.
The basic echo server from src/examples/minimal.ml:
simple_httpd/examples/minimal.ml
Lines 1 to 27 in c7dde27
>>> dune exec -- examples/minimal.exe -c 1000 --log-requests 0 &
[4] 341587
listening on http://127.0.0.1:8080bs: 0)
>>> curl -X GET http://localhost:8080/echo --data "howdy y'all"
echo:
{meth=GET; host=localhost:8080;
headers=[Content-Type: application/x-www-form-urlencoded
Content-Length: 11
Accept: */*
User-Agent: curl/7.88.1
Host: localhost:8080];
path="/echo"; body="howdy y'all"; path_components=["echo"]; query=[]}
>>> wrk -t5 -c500 -d5 http://localhost:8080/echo
Running 5s test @ http://localhost:8080/echo
5 threads and 500 connections
Thread Stats Avg Stdev Max +/- Stdev
Latency 4.09ms 2.80ms 48.13ms 85.33%
Req/Sec 24.79k 2.26k 30.04k 70.80%
618815 requests in 5.05s, 100.92MB read
Requests/sec: 122487.56
Transfer/sec: 19.98MBUsing wrk, we see the good performances: more than 120000 requests per seconds, on my small 6 core laptop, with the 5 client threads and 5 server threads running on the same CPU.
Similar to python -m http.server, a simple program http_of_dir is provided.
It serves files from the current directory.
$ http_of_dir . -p 8080 &
$ curl -X GET http://localhost:8080
...
<html list of current dir>
...
The program http_of_dir relies on the module Simple_httpd_dir, which
can serve directories, as well as virtual file systems.
In 'examples/dune', we produce an OCaml module vfs.ml using
the program vfs_pack. This module contains a VFS (virtual file
system) which can be served as if it were an actual directory.
The dune rule:
(rule
(targets vfs.ml)
(deps (source_tree files) (:out test_output.txt.expected))
(enabled_if (= %{system} "linux"))
(action (run ../src/bin/vfs_pack.exe -o %{targets}
--mirror=files/
--file=test_out.txt,%{out}
--url=example_dot_com,http://example.com)))The code to serve the VFS from vfs.ml is as follows:
…
Simple_httpd_dir.add_vfs server
~config:(Simple_httpd_dir.config ~download:true
~dir_behavior:Simple_httpd_dir.Index_or_lists ())
~vfs:Vfs.vfs ~prefix:"vfs";
…it allows downloading the files, and listing directories.
If a directory contains index.html then this will be served
instead of listing the content.
An html file named foo.chaml will be served as foo.html but may use some OCaml
code embedded inside the html, a bit like php, but statically compiled. Here
is an example of chaml file:
Why not? If you just want a super basic local server (perhaps for exposing data from a local demon, like Cups or Syncthing do), no need for a ton of dependencies or high scalability libraries.
This is a fork of the original tiny_httpd by Simon Cruanes to experiment with OCaml 5 domains and effects. It remains simple and seems to work very well.
The plan is to try to maximise features while keeping it simple.
Use cases might include:
- serve content directly from a static blog generator;
- provide a web UI to some tool (like CUPS and syncthing do);
- implement a basic monitoring page for a service;
- provide a simple json API for a service, on top of http;
- use
http_of_dirto serve odoc-generated docs or some assets directory. - experimenting with domains and effects
See https://raffalli.eu/simple_httpd/simple_httpd
MIT.