Provides an opinionated, but rational, set of mix tools for managing version numbers for your Elixir project using the following scheme.
Major.Minor.Patch
You decide what each version number means, whether it's semantic versionning, Spec-ulation from Rich Hickey, or any other scheme.
@deps [
{:version_tasks, "~> 0.12.0"}
]A set of Mix Tasks for managing your version numbers. This is best used in conjunction with your release strategy, and so we also provide some opionated generated code / bash scripts to support you in your quest for towards continuous deployment.
To get a list of all available mix tasks (in case the docs might be stale), you can run
mix version
Below is a bit more information about the available tasks and how to use them.
The following tasks just report back information about your project without actually changing it.
Here's the basic usage
MIX_QUIET=1 mix version.<current|inc|next|tag|up>
Notice the MIX_QUIET=1, can sometimes be important if you are
using this within a dev environment where additional debugging
output might be included
The examples below will omit the MIX_QUIET=1for brevity, and as
it isn't strictly required.
To retrieve the current version of your application, run
mix version.current
The call back returns the current version of your project, for example
1.2.3
This is based on your mix.exs file.
To retrieve the next version of your application, run
mix version.next <major|minor|patch>?
The default increment step is patch, here are a few examples from the version above
mix version.next
1.2.4
mix version.next patch
1.2.4
mix version.next minor
1.3.0
mix version.next major
2.0.0
After an upgrade, you might want to trigger additional actions, such as run tests
create a release and deploy an update. You can ask for the name of the version using:
# For anything like X.0.0, that's a major release
mix version.name
major
# For anything like X.Y.0, that's a minor release
mix version.name
minor
# For all other releases, like X.Y.Z, that's a patch release
mix version.name
patch
This is the same as mix version.name with the exception that it will only return the
version name IF the last commit was a version commit, which looks like vX.Y.Z.
# If the last commit was `Doing something great`, then it returns nothing
mix version.is_release
# However, if the last commit was `v1.2.0`, then it returns `minor`
mix version.is_release
minor
To help us trigger when a release has been requested, we are exposing the last commit
message, this delegates to git and simply calls (git log --format=%B -n 1 HEAD)
# Grab the last commit
mix version.last_commit
Simplify the database backup, to make restore easier
# Hey, this commmit looks like a new release was commit
mix version.last_commit
v1.4.5
You can also expose just the major number, or minor number or patch number.
This would be useful for scriping where you wanted to join the numbers with
a dash, and your bash is not up to the par, so you call it out individually.
# Let's assume `mix version.current` is 1.2.3
mix version.major
1
mix version.minor
2
mix version.patch
3
The following tasks will edit your local files, but will not commit or push any of those changes.
Increment your project to the next version, this will update your mix.exs AND your README.md file.
mix version.inc <major|minor|patch>?
Your mix.exs MUST HAVE a variable named as follows for this to work
@version "1.2.3"
This is the default is based on a template from Dave Thomas with his mix gen <template>
alternative to mix new. A video explaining mix gen and mix template
And your README.md SHOULD HAVE an installation section as follows:
@deps [
{:your_app_name, "~> 0.12.0"} # <-- that's the important line to have
]
These next set of tasks will commit to your local repository and/or push to your local repository.
Upgrade the version number on your project and commit the changes files to your local git repository.
mix version.up <major|minor|patch>?
Following the version.up, we can tag and push the changes to your remote branch.
mix version.tag
This will tag your repo with vMajor.Minor.Patch and push the tag to your remote branch.
Undoes the effects of a mix version.tag
mix version.untag
This will remove the tag in your repo with vMajor.Minor.Patch and push the delete to the remote branch as well.
Before you can use this task, you need to ensure you have registered with hex directly, you can do this from the command line with
mix hex.user register
If you already have an account, but you haven't authenticated yourself on this device, then you can run
mix hex.user auth
Once you have an account, and are authorized, a upgrade script can look like. Here's a script to upgrade your package on hex
mix do version.up <major|minor|patch>, version.tag && \
mix test && \
mix hex.publish
I usually drop this in a ./bin/hexup bash script in my project,
as mix test needs the MIX_ENV=test, and you cannot combine
updating the version number with hex.publish as it won't pick
up the new version you just created.
You can configure an automatic push to hex, but to do that you cannot have a passphrase associated with your key.
mix hex.user passphrase
And then you install a git hook which will install a post-commit hook
mix githooks.hexup
If you want to keep your passphrase, then it will be stored in plaintext inside your .git/hooks directory, simply call
mix githooks.hexup <passphrase>
Your passphrase should be different then your hex.pm password.
You can configure an automatic deploy (as defined in a script
./bin/deploy using a git post-commit hook. To install it, run
mix githooks.deploy
This will install a .git/hooks/post-commit hook that when a new
version is detected will run a ./bin/deploy script. You can specify
your own script by running
mix githooks.deploy bin/customdeploy
If the deploy file does not exist, then we will create one for you, but it won't do much, basically just
mix test && \
mix version.tag
When creating releases (aka mix release), there are a few scripts that
are handy to have around to start a console, or upgrade the release, etc.
To install these scripts into your project, run
mix version.bin.release <release_dir>?
You can provide an optional <release_dir> (defaults to /src/{appname}rel)
which knows how to grab your created release and put it in a git repo. If
you don't know or don't care for this, then ignore it.
This will create the following files:
./bin/package/prerelease # Fetch dependencies, compile, build and digest assets
./bin/package/release # Generate an "upgradeable" ERTS release
./bin/package/retain # Commit your release to a separate repo
./bin/run/rel # run any other `release` task available
./bin/run/launch # Start you app (if stopped), upgrade if running
./bin/run/debug # Start you app from code in an iEX shell
It will also create two custom commands, and place them in
./rel/commands/clear_cache # Clear the phoenix cache to support hot code swapping
./rel/commands/migrate # Migrate your ecto database
To make these available, you will need to follow the instructions to add them
to your rel/config.exs file. It will look similar to:
release :<your appname> do
...
# Add these to your :commands
set commands: [
"clear_cache": "rel/commands/clear_cache",
"migrate": "rel/commands/migrate"
]
end
Once available, you will be able to call them directly, such as
./bin/<appname> clear_cache
./bin/<appname> migrate
Please note that Phoenix (at present), was not reloading the re-compiled static
assets on an upgrade, so we also write a &<AppModule>.ReleaseTasks.clear_cache/0 to
deal with ensuring that javascript and CSS are properly available.
./lib/<appname>/release_tasks.ex
You will need to commit these files to you project. If you edit them, please let me ([email protected]) as the changes might be relevant to others.
This will create a &<AppModule>.FeatureFlags GenServer, which is really just a
simple map of enabled or disabled atoms.
To install these scripts into your project, run
mix version.bin.ff
To enable a feature flag, say deploying, run
<AppModule>.FeatureFlags.enable(:deploying)
To disable that flag, run
<AppModule>.FeatureFlags.disable(:deploying)
In your code, you can then make decisions based on those flags
<%= if <AppModule>.FeatureFlags.enabled?(:deploying) do %>
<p>We are deploying updates to server you better</p>
<% end %>
We will also expose that functionality within your release. It will create two custom scripts
./rel/commands/enable <flag> # Enable the provided feature <flag>
./rel/commands/disable <flag> # Disable the provided feature <flag>
To make these available, you will need to follow the instructions to add them
to your rel/config.exs file. It will look similar to:
release :<your appname> do
...
# Add these to your :commands
set commands: [
"enable": "rel/commands/enable",
"disable": "rel/commands/disable"
]
end
You will need to commit these files to you project. If you edit them, please let me ([email protected]) as the changes might be relevant to others.
This will add helper scripts to bin/db for backing up and restoring your database. Very opinionated, and only works for postgres.
To install these scripts into your project, run
mix version.bin.db <backup_root> <dbname>
The default backup_root will be <appname>backup and your dbname will be
your <appname>. Overwrite at your leisure.
The script will create the following files:
./bin/db/backup # Backup your database
./bin/db/restore # Restore your database
You will need to commit these files to you project. If you edit them, please let me ([email protected]) as the changes might be relevant to others.
MIT License
Created: 2017-07-05Z