pico-build is the world's smallest, yet featureful, three-environment (dev, stage and prod) build and deploy system.
It is intended for individual programmers and small teams who want to test and deploy to three environments, but don't want to spend time setting up (and patching) Jenkins or building a CI/CD pipeline.
pico-build is a Makefile of only about 10 active executable lines that:
- is a command-line tool that does input validation on the deploy environment choice
- updates the
dev/folder using your version control system - exports your tested version in
dev/tostage/andprod/folders - has a distribution target, dist, for multi-server deploys using rsync or bittorrent
- can be setup in a minute or two
- for more information about
make, see the GNU make homepage - source code: Makefile
Also, by running pico-build using sudo or root, pico-build can be used to secure your web server file ownership and permissions.
Normally, most make dependencies are file timestamps. However, the pico-build Makefile does not use timestamps,
only phony targets and the repo version. Thus make is used like a Lisp or Prolog functional programming rules engine. The bash exit command is used to stop early, like Prolog's cut.
pico-build is a locally-centric build system, but that's often not a limitation since it can fetch from remote version control systems, and deploy to remote clusters as needed.
Functional programming with GNU make
usage: make [help|check|dev|stage|prod|dist|all]
- choose your
gitorsvnrepo and configure your login environment to do a successful password-lessgit pull(or equivalent command) from the terminal - add
.current_versionto your.gitignorefile and commit.gitignore cdto your build home directory which is the parent of yourdev/,stage/andprod/folders and do an initialgit cloneto thedevfoldercpthepico-buildMakefileto the home directory and configure as needed. Change the permissions onMakefileso that unauthorized users cannot modify it, usually withchown root:root Makefile; chmod 644 Makefile- do
make checkto do an initial test of your configuration - run
makefor each environment:
make dev
# run your tests until they pass ... then ...
make stage
make prod
# for multi-server deployments, you can use rsync or bittorrent ...
make dist
After you have tested the above, you can optionally run all of the deploy steps sequentially with this command (not recommended unless you have great tests):
make all
pico-build has several advantages over Jenkins:
- the easiest way to get started learning about three-environment builds
- no waiting for your deploy job in a central build queue
- can itself be committed to version control
- secure
- no additional RAM or disk space required
- can be distributed widely to clusters or end-users
- requires no licensing or support (if you do need commercial support, see the FAQ below)
- portable - runs on all commonly-used operating systems
- many release engineers and QA staff are already familiar with
make - less is more!
pico-build is especially useful for Operations teams who work on master or HEAD branches and often don't get around to setting up deploy environments for their own team, or who don't want to cut holes in firewalls (or peer across AWS VPC boundaries.)
It can also be used to develop and/or create secure deploy processes.
pico-build will work on any operating system that supports make. It has been tested on CentOS/Redhat and Mac OS X which both use GNU make. It will probably work on Windows under Cygwin or Microsoft Windows Subsystem for Linux (WSL).
To install make on CentOS or Redhat, type sudo yum install make.
Question: Why is the name pico-build?
Answer: Originally I wanted to use nano-build, but nano-os already has a build system.
Question: It's very small, but aren't you cheating by using the make program?
Answer: Yes. Yes I am. Shamelessly.
Question: In fact, pico-build is so small, how can you justify creating a Github project based on a 10-line Makefile?
Answer: Although the Makefile is very short, pico-build delivers on what it promises. A previous version in bash took about a week to design and test, and required much more code to get the same result.
Question: Can cron run pico-build?
Answer: Yes, but you'll need to do this:
- ensure
makeandcroncan still read your version control credentials, and can findmake,gitandtar, and set the path, likePATH=/bin:/usr/bin. - quiet or redirect output to a file to reduce the
cronemail notifications - either set the cwd or do
make -f /path/to/the/pico-build/Makefile action.
Question: How does the pico-build Makefile work?
Answer: pico-build uses basic make functionality:
help,check,dev,stage,prod,distandallaremaketargets, or actions. When you saymake dev,devis the target rule and$@is assigned 'dev'.DEBUGtoggles$(DISP)to @ or blank to controlmake's display of command execution.- You can do
make -n, which is the dry-run mode, to see what commands thatmakewould execute.
Question: I'm new to version control but I want to use pico-build. Which version control program should I use?
Answer: You should use the same version control program that your company or friends use. Github uses git. Otherwise, svn is my favorite and is the easiest to learn because of its clear command structure. Set the VC_PRODUCT variable in Makefile to select either git or svn.
Question: Does pico-build check for changes in dev/ before doing pointless stage and prod deploys?
Answer: Yes, this is handled by comparing the content of dev/.current_version against the same filename in stage/ or prod/.
Question: I copypasta'ed your Makefile and it doesn't seem to work for me. Why?
Answer: make requires tab characters before bash commands, so check if you accidently converted the tabs to spaces. (vim will helpfully show red lines if you do that.) Otherwise run with make -n for a dry run.
Question: Is there commercial support available for pico-build?
Answer: Yes, please contact the author for paid Devops design and/or support via the blog contact form.
Copyright James Briggs, USA 2018
MIT License