master
branchmaster
branchKetrew applications (ie. server or client), use a JSON format for configuration (it is possible to generate that file from OCaml or even to avoid all configuration files by creating your own ad-hoc command-line application).
The configuration file can contain one or more named "profiles".
Ketrew finds it's configuration file by successively checking:
--configuration-file
(alias: -C
)KETREW_CONFIGURATION
and KETREW_CONFIG
,KETREW_ROOT
or the default
$HOME/.ketrew/
), Ketrew will try to access the files:configuration.json
,configuration.ml
, andconfiguration.sh
Given the extension of the filename Ketrew will read the configuration differently:
.json
→ will read and parse the file;.ml
→ will execute ocaml <file>
, collect stdout
, and parse it (as JSON);.sh
→ will execute ./<file>
, collect stdout
, and parse it (as JSON).Other file extension are considered undefined behavior for future use.
After parsing the configuration file, Ketrew will select a profile by name:
-P
/--configuration-profile
,KETREW_PROFILE
,"default"
.The command ketrew initialize
can generate a configuration file (among other
things); see ketrew init --help
.
A configuration file configuration.ml
(that Ketrew will execute through
OCaml
) would look like:
#use "topfind" #thread #require "ketrew" open Ketrew.Configuration let debug_level = 2 (* `debug-level`: integer specifying the amount of verbose messages: `0`: none, `1`: verbose, `2`: very verbose. *) (* Plugins to load: *) let plugins = [ `OCamlfind "lwt.react"; `Compiled "/path/to/some/ketrew_plugin.cmxs"; ] (* User-Interface preferences: *) let explorer = explorer ~request_targets_ids:(`Younger_than (`Days 1.5)) ~targets_per_page:5 ~targets_to_prefetch:10 () let ui = ui ~with_color:true ~explorer ~with_cbreak:true () (* A function that given a boolean value creates a “server configuration” that detaches or not from the shell. *) let my_servers daemon = server ~ui ~engine:(engine ~database_parameters:"/path/to/database-client-server" ()) ~authorized_tokens:[ authorized_tokens_path "/path/to/authorized-tokens"; authorized_token ~name:"The-inline-one" "inlinetoken"; ] ~return_error_messages:true ~log_path:"/path/to/logs-of-server.txt" ~daemon ~command_pipe:"/path/to/command.pipe" (`Tls ("/path/to/cert.pem", "/path/to/key.pem", 8443)) (* We put together 4 profiles in this configuration and “output” them (literally, as Json, to `stdout`). `debug_level`, `plugins`, and `ui` are shared between configurations. *) let () = output [ profile "standalone" (create ~debug_level ~plugins (standalone ~ui () ~engine:(engine ~database_parameters:"/path/to/database-standalone" ()))); profile "daemon" (create ~debug_level ~plugins (my_servers true)); profile "server" (create ~debug_level ~plugins (my_servers false)); profile "client" (create ~debug_level ~plugins ( client ~ui ~token:"nekot" "https://127.0.0.1:8443" )); ]
You may run ocaml configuration.ml
to see the equivalent Json.
Creating a test environment (make test-env
, cf. developer
docs) generates a pretty complex configuration
file.
To build configurations refer to the API of
the Ketrew.Configuration
module.
As shown above, the idea is to call the function Ketrew.Configuration.output
with results of the function Ketrew.Configuration.profile
, themselves created
thanks to the function Ketrew.Configuration.create
, etc.
One can always test their configuration with:
ketrew print-configuration