=head1 NAME

Dist::Milla::Tutorial - Milla HOW TO

=head1 WORKFLOW

See also L<the
screencast|http://weblog.bulknews.net/post/46471116934/introducing-milla>
explaining how to setup and create a new distribution with Milla.

=head2 Setup

Install Milla and setup your profile.

  > cpanm Dist::Milla
  > milla setup

Setup command will ask you a simple question to make a basic
profile. If you already have set up C<dzil> before, this is common and
you can skip the process.

=head2 Making a new distribution

Now it's time to make a new distribution.

  > milla new Dist-Name
  > cd Dist-Name

At this point, you will have a really simple C<Dist-Name> directory
that contains your module file with as minimum boilerplate as
possible.

It is recommended to track your repository under git as soon as
possible, even before releasing to CPAN.

  # git is already initialized and files are added for you
  > git commit -m "initial commit"

Now start writing your code, edit the docs, tests and manage CPAN
dependencies with L<cpanfile>.

  > $EDITOR lib/Dist/Name.pm t/dist-name.t cpanfile

You can test your code with a simple C<prove -l t>.

For the first time build only, you can make a test build to get some
boilerplate you want to keep in the git repository so that your github
repository looks great with README, as well as installable from git
using C<cpanm> or testable with C<Travis CI>.

  > milla build
  > git add Build.PL META.json README.md && git commit -m "git stuff"

=head2 Making the first release

When you get confident and it's about time to ship to CPAN, use the
test and release command. Before doing so, make sure your git
directory is not dirty i.e. all changes are committed.

  > git commit -a -m "Done initial version"

Milla assumes you have a git remote setup so that you can push all
your changes to. I recommend you to use either
L<hub gem|https://rubygems.org/gems/hub> or L<App::ph> to create a new
github repository.

  # Use hub rubygem
  > hub create

  # Use App::ph
  > ph import

Now, make sure you have C<Changes> file ready and have a new entry
under C<{{$NEXT}}>, which will be expanded to the next version of your
module.

  > $EDITOR Changes
  > milla test
  > milla release

And your first release is done. The release is tagged on git and all
the changes atomatically made are committed to git as well.

If this is your first conversion to Milla and want to make sure you're
not going to mess CPAN with a bad archive when something goes wrong,
you can run the release command with C<FAKE_RELEASE> environment
variable. This will run all the other release process, except the
UploadToCPAN step.

  > FAKE_RELEASE=1 milla release

Wait for PAUSE processing it and your module showing up on MetaCPAN in
a few minutes. Congratulations!

=head2 Making a maintainance release

You have new features, bugs, pull requests and get ready to make a next
version of your module. Great, making a new release is equally easy.

First, make sure all your code has been committed to git and there are
no dirty files in the working directory.

Then make sure to edit C<Changes> file and contain entries for the
next release under C<{{$NEXT}}>. You don't need to commit the change
to the C<Changes> file, yet.

Now, make a release!

  > milla test
  > milla release

The release command will automatically bump the version for you - if
you have C<0.10>, the next version will be C<0.11> by default, but you
will be prompted to confirm that version in case you need a major
bump.

This will update C<Changes>, C<META.json> and bump C<$VERSION> in your
main module. These changes made by Milla will be automatically
committed, tagged and pushed to the remote.

=head1 MIGRATING

This section describes how to migrate your current authoring process
to Milla.

=head2 Automatic migration

Sister project L<Minilla> provides a wonderful conversion tool to
migrate your existing distribution to Milla, which happens to be
compatible to Milla as well.

As an author of Milla, I use Minilla's migrate command all the time to
get the most of conversion done.

To run Minilla migration tool, install L<Minilla> from CPAN and run
the C<minil migrate> command in the project root.

    > cpanm Minilla
    > minil migrate && git rm -f minil.toml

The command will output the migration process, and C<git add> all the
changed files for you. Once the migrate command is complete, all you
have to do is to create C<dist.ini> file, and run the build command.

    > echo "[@Milla]" > dist.ini
    > milla build

If one of the command has failed, you might need to reset to the
original state and run the manual migration. Read on to see how to do
that.

=head2 Manually migrating from other build tools

=head3 Module Dependencies to cpanfile

First, move the prereq declaration from C<Makefile.PL> or C<Build.PL>
to C<cpanfile>.

The easiest way to convert existing dependencies to cpanfile is to use
the command line tool C<mymeta-cpanfile>, which is installed with
L<Module::CPANfile>. Run the configuration with C<Makefile.PL> for the
one last time, then run the C<mymeta-cpanfile> command:

  > perl Makefile.PL
  > mymeta-cpanfile --no-configure
  requires 'DBI', '1.000';
  
  on test => sub {
      requires 'Test::More', '0.86';
  }
  
  ...

You can redirect the output to C<cpanfile> if you like. It is
important to pass C<--no-configure> option here, since otherwise
modules like ExtUtils::MakeMaker will be included. It is not required
with Milla setup, since Milla knows which configuration tool
(installer) to use and include them in META files upon the
releases. You can leave that out from the C<cpanfile>.

If you decide to manually construct a new cpanfile, the format is mostly
compatible to L<Module::Install>'s requirement DSL.

  # Makefile.PL
  test_requires 'Test::More', 0.90;
  requires 'Plack', '1.000';

becomes:

  # cpanfile
  test_requires 'Test::More', 0.90;
  requires 'Plack', '1.000';

which is exactly the same. If you use L<Module::Build> or
L<ExtUtils::MakeMaker>, that will be more manual process, but
basically the same thing. See L<cpanfile> for the available syntax.

=head3 Remove boilerplate 

Next, remove unnecessary boilerplate files.

  > git rm {Makefile,Build}.PL MANIFEST MANIFEST.SKIP README .shipit

=head3 Create a new ini and edit configurations

Next, create a new C<dist.ini> with the following two lines:

  name = Dist-Name
  [@Milla]

the C<name = > line is optional.

If your work directory is named C<Dist-Name>, Milla will be able to
figure out that is your distribution name, so you can omit that line.

Next, Edit C<.gitignore> and add the following lines:

  /Dist-Name-*
  /.build
  !META.json

You're almost done, and your directory will look like:

  cpanfile
  dist.ini
  lib/Dist/Name.pm
  t/...

C<git add> the newly created files and commit it.

=head3 Make a new build

Now you're ready to make the first build.

  > milla build

and if it was successful, you get a build in a directory called
C<Dist-Name-v0.1.0> under your current directory. They can be later
removed with C<milla clean> command.

Also, new C<Build.PL>, C<META.json> and C<README.md> are added in your
working directory for git-friendliness. C<git add> them and commit it.

  > git add Build.PL META.json README.md && git commit -m "git stuff"

Now you're ready to roll a new release with Milla. Before doing so,
convert your C<Changes> file format a little bit, and make sure you
have a following header in the top:

  {{$NEXT}}
          - Change log entry for the next version

The C<{{$NEXT}}> is a template variable that gets replaced with the
version and date string, when you make a next release. This is almost
the I<only> change you're required to make in your code base.

Now, run the release command:

  > milla release

to make a new release, in the same way described above for a new Milla
setup. You can set C<FAKE_RELEASE> environment variable if this is
your first conversion and want to double check what happens, before
uploading to CPAN.

When this is not your first release, the version number gets
automatically bumped by Milla, but you will be prompted if that is
exactly the version you want, and if you want a major version up, you
can specify to do so.

=head1 AUTHOR

Tatsuhiko Miyagawa

=head1 SEE ALSO

L<Dist::Milla>

=cut