fukamachi / shelly

[OBSOLETE] Use Roswell instead.
http://shlyfile.org
BSD 2-Clause "Simplified" License
63 stars 6 forks source link

Shelly - Make Common Lisp shell-friendly.

Announcement: Release of v0.7.5

Perl5 isn't required to use Shelly from v0.7.0 because I rewrote it in shell script.

You can install the latest version by this command:

$ curl -L http://shlyfile.org/shly | /bin/sh

If you already have an older version of it, you can upgrade it by this command:

$ shly install --version latest

Usage

Usage: shly [option,..] <command> [arg1,arg2..]

Examples:

$ shly asdf:test-system :clack
$ shly ql:update-all-dists --prompt nil
$ shly -Lclack clackup /path/to/project/app.lisp
$ shly -Lclack clack.app.directory:start-server
$ shly -Ldrakma http-request http://www.hatena.com/
$ shly -Lcl-project make-project /path/to/myapp/ --description "My sample app." --author "Eitaro Fukamachi"

Description

Shelly enables you to execute Common Lisp functions like a shell command. And it also can be used as a Make-like build-tool.

Shelly has the following features:

Warning: This software is still ALPHA quality. The APIs will be likely to change.

Why use it?

In Common Lisp world, most libraries and applications are designed to run on REPL. It is convenient for interactive development, but it would be an obstacle in some cases.

For example, Common Lisp isn't good for writing a small script. No common way to write it in a portable way. Parsing command-line arguments is really annoying. And, the startup time would be very slow, especially when it uses some libraries.

Shelly solves these problems by providing a shell-friendly interface of Common Lisp. If your application need to run with Cron, Supervisord or other CUI applications, this may help you.

1. Implementation independent

Shelly should work fine with one of SBCL, Clozure CL, Allegro CL, ABCL, GNU CLISP and ECL.

2. Function as a shell command

Shelly treats general functions as its sub-commands, so you don't even need to write a script in most cases.

(in-package :myapp)

(defun do-something (&key verbose)
  ;; Do something.
  )

$ shly myapp:do-something --verbose t

Command-line options and arguments will be delivered to a function.

3. Fast startup

Shelly reduces the startup time by storing a Lisp core image. In a simple case, the execution is about 33 times faster than CIM's cl command and even 25 times faster than SBCL (with Quicklisp) at the maximum.

# Uses SBCL v1.2.1, Shelly v0.7.0
$ time shly + 1 2
3
shly + 1 2  0.02s user 0.03s system 102% cpu 0.047 total

# CIM v1.0.0
$ time cl --eval '(princ (+ 1 2))'
3
cl --eval '(princ (+ 1 2))'  0.67s user 0.11s system 96% cpu 0.805 total

# SBCL with Quicklisp
$ time sbcl --noinform --eval '(princ (+ 1 2))' --eval '(quit)'
3
sbcl --noinform --eval '(princ (+ 1 2))' --eval '(quit)'  0.51s user 0.09s system 99% cpu 0.606 total

# SBCL without Quicklisp
$ time sbcl --noinform --no-userinit --eval '(princ (+ 1 2))' --eval '(quit)'
3
sbcl --noinform --no-userinit --eval '(princ (+ 1 2))' --eval '(quit)'  0.00s user 0.01s system 89% cpu 0.012 total

How does it work

Shelly provides a shell script "shly". It takes some options, a command, and arguments for the command.

Usage: shly [option,..] <command> [arg1,arg2..]

In this example, ql:system-apropos would be the command and web would be an argument.

;; Same as (ql:system-apropos "web")
$ shly ql:system-apropos web
#<SYSTEM bknr.modules / bknr-web-20140616-git / quicklisp 2014-06-16>
#<SYSTEM bknr.web / bknr-web-20140616-git / quicklisp 2014-06-16>
#<SYSTEM cl-web-crawler / cl-web-crawler-20130128-svn / quicklisp 2014-06-16>
#<SYSTEM cl-webdav / cl-webdav-0.2.1 / quicklisp 2014-06-16>
#<SYSTEM crane-web / crane-20140616-git / quicklisp 2014-06-16>
#<SYSTEM hh-web / hh-web-20140616-git / quicklisp 2014-06-16>
...

If an argument starts with ":", it would be converted into a keyword.

;; Same as (asdf:system-source-file :hunchentoot).
$ shly asdf:system-source-file :hunchentoot
#P"/Users/nitro_idiot/quicklisp/dists/quicklisp/software/hunchentoot-1.2.21/hunchentoot.asd"

If an argument starts with "--", it also would be a keyword. This is just like common shell commands.

;; Same as (ql:update-all-dists :prompt nil)
$ shly ql:update-all-dists --prompt nil

If the command is imported to COMMON-LISP-USER package, you can omit the package prefix.

$ shly list 1 2 3
(1 2 3)

Or, if the package name is the same as the system name that is loaded by -L option, you can also omit it.

$ shly -Ldrakma http-request http://www.hatena.com/
$ shly -Lcl-project make-project /path/to/myapp/ --description "My sample app." --author "Eitaro Fukamachi"

As a replacement of Makefile

Shelly loads a local file which is named shlyfile.lisp if it exists. You can define project specific commands by writing functions in it. This is just like a "Makefile" in Common Lisp.

;; shlyfile.lisp
(defun test ()
  (asdf:test-system :your-app))

(defun build ()
  ;; Somthing to build your app.
  )

Then, shly test and shly build would be available only in the directory.

$ shly test
$ shly build

Shelly also loads ~/.shelly/shlyfile.lisp every time if it exists. If you have some commands you'd like to use everywhere, put them into that file.

Requirements

If you've installed CIM, Shelly will use its setting.

Installation

Installing the latest version

Currently, the install script always installs the latest version because the stable version requires Perl5 and I suppose you may not expect it.

$ curl -L http://shlyfile.org/shly | /bin/sh

You can also install from the source code.

$ git clone https://github.com/fukamachi/shelly.git
$ cd shelly
$ SHELLY_PATH=. bin/shly install

As "shelly" directory will be copied to ~/.shelly, you don't need the cloned repository after installation.

Installing to other than ~/.shelly

If you want to install Shelly to the different location, set SHELLY_HOME to the directory path.

$ curl -L http://shlyfile.org/shly | SHELLY_HOME=~/.shly /bin/sh

System-Wide installation

If --global t is specified to install command, it would be installed under /usr/local.

$ curl -L http://shlyfile.org/shly | /bin/sh -s install --global t

Configuration

If you use bash, zsh, csh or tcsh, the initialization code will be appended into your .*rc file automatically.

Otherwise, set SHELLY_HOME and add ~/.shelly/bin to PATH manually.

SHELLY_HOME="$HOME/.shelly"
[ -s "$SHELLY_HOME/lib/shelly/init.sh" ] && . "$SHELLY_HOME/lib/shelly/init.sh"

Upgrading

shly upgrade

Uninstalling

$ shly uninstall
$ rm -rf ~/.shelly

Built-in Commands

help [&optional command]

Show the usage of the specified command.

$ shly help ql:quickload
Usage: ql:quickload (systems &key verbose prompt explain
                     (verbose *quickload-verbose*) (prompt *quickload-prompt*)
                     &allow-other-keys)
    Load SYSTEMS the quicklisp way. SYSTEMS is a designator for a list
       of things to be loaded.

If command is not specified, it shows all available commands. This is the same as shly --help.

$ shly help

install [&key version global directory]

Install Shelly into your environment under "~/.shelly". You can install a specific version by using "--version".

$ shly install --version v0.6.1

If --directory is specified, it would be installed to the directory.

$ shly install --directory ~/.local

If --global is specified with non-NIL value, it would be installed to /usr/local/ (Same as --directory /usr/local).

$ shly install --global t

uninstall [&key directory]

Uninstall Shelly.

$ shly uninstall

available-versions

Show all the possible Shelly versions.

$ shly available-versions

dump-core

Dump Lisp core image file for faster startup.

$ shly dump-core

rm-core

Remove saved core image file which created by dump-core.

$ shly rm-core

local-dump-core [&rest systems] (Experimental)

Dump Lisp core image file to the current directory. This command takes system names to be included in the core.

$ shly local-dump-core :myapp

install-command [package-or-function-name]

Make an executable file under SHELLY_HOME/bin/.

# Making a function executable.
$ shly -Lclack install-command clack:clackup
# Same as (clack:clackup #P"/path/to/app.lisp")
$ clackup /path/to/app.lisp

# Making a package executable.
$ shly -Lclack install-command clack.app.directory
# Same as (clack.app.directory:start-server :port 50032)
$ clack.app.directory start-server --port 50032

History (roughly)

v0.8.6 (Oct 27, 2014)

v0.8.5 (Oct 3, 2014)

v0.8.1 (Aug 4, 2014)

v0.8.0 (Aug 2, 2014)

v0.7.0 (July 15, 2014)

v0.6.0 (Mar 31, 2014)

v0.5.8 (Nov 12, 2013)

v0.5.0 (Aug 29, 2013)

v0.4.0 (Aug 26, 2013)

Note: Conversion rules

Number

$ shly type-of 24
(INTEGER 0 4611686018427387903)
$ shly type-of 3.141592653589793d0
DOUBLE-FLOAT

Cons

$ shly type-of '(list 1 2 3)'
CONS

Boolean

If it is "t" or "nil", it would be T or NIL.

$ shly type-of t
BOOLEAN
$ shly type-of nil
NULL

Keyword

If it starts with ":" or "--", it would be a keyword.

$ shly type-of :web
KEYWORD
$ shly type-of --verbose

Pathname

If the same name file exists, it would be a pathname. Otherwise, it would be just a string.

$ shly type-of test.lisp
(SIMPLE-ARRAY CHARACTER (9))
$ touch test.lisp
$ shly type-of test.lisp
PATHNAME

Symbol

If the same name symbol exists, it would be a symbol.

$ shly pi
DOUBLE-FLOAT
$ shly asdf:\*central-registry\*
CONS

String

Otherwise, it would be treated as a string.

$ shly common-lisp-is-great
(SIMPLE-ARRAY CHARACTER (20))

Copyright

Copyright (c) 2012-2014 Eitaro Fukamachi and contributors.

License

Licensed under the BSD (2-Clause) License. See LICENSE.