Manpages - Alien_Build_Manual_PluginAuthor.3pm

$meta->register_hook( probe => sub { my($build) = @_; return system if …; # system install return share; # otherwise }); $meta->register_hook( probe => [ $command ] );

This hook should return the string system if the operating system provides the library or tool. It should return share otherwise.

You can also use a command that returns true when the tool or library is available. For example for use with pkg-config:

$meta->register_hook( probe => [ %{pkgconf} –exists libfoo ] );

Or if you needed a minimum version:

$meta->register_hook( probe => [ %{pkgconf} –atleast-version=1.00 libfoo ] );

Note that this hook SHOULD NOT gather system properties, such as cflags, libs, versions, etc, because the probe hook will be skipped in the event the environment variable ALIEN_INSTALL_TYPE is set. The detection of these properties should instead be done by the gather_system hook, below.

$meta->register_hook( gather_system => sub { my($build) = @_; $build->runtime_prop->{cflags} = …; $build->runtime_prop->{libs} = …; $build->runtime_prop->{version} = …; });

This hook is called for a system install to determine the properties necessary for using the library or tool. These properties should be stored in the runtime_prop hash as shown above. Typical properties that are needed for libraries are cflags and libs. If at all possible you should also try to determine the version of the library or tool.

$meta->register_hook( download => sub { my($build) = @_; … });

This hook is used to download from the internet the source. Either as an archive (like tar, zip, etc), or as a directory of files (git clone, etc). When the hook is called, the current working directory will be a new empty directory, so you can save the download to the current directory. If you store a single file in the directory, Alien::Build will assume that it is an archive, which will be processed by the extract hook below. If you store multiple files, Alien::Build will assume the current directory is the source root. If no files are stored at all, an exception with an appropriate diagnostic will be thrown.

Note: If you register this hook, then the fetch, decode and prefer hooks will NOT be called.

package Alien::Build::Plugin::MyPlugin; use strict; use warnings; use Alien::Build::Plugin; use Carp (); has +url => sub { Carp::croak “url is required property” }; sub init { my($self, $meta) = @_; $meta->register_hook( fetch => sub { my($build, $url, %options) = @_; … } } 1;

Used to fetch a resource. The first time it will be called without an argument (or with $url set to undef, so the configuration used to find the resource should be specified by the plugin’s properties. On subsequent calls the first argument will be a URL.

The %options hash may contain these options:

http_headers

HTTP request headers, if an appropriate protocol is being used. The headers are provided as an array reference of key/value pairs, which allows for duplicate header keys with multiple values. If a non-HTTP protocol is used, or if the plugin cannot otherwise send HTTP request headers, the plugin SHOULD issue a warning using the $build->log method, but because this option wasn’t part of the original spec, the plugin MAY no issue that warning while ignoring it.

Note that versions of Alien::Build prior to 2.39 did not pass the options hash into the fetch plugin.

Normally the first fetch will be to either a file or a directory listing. If it is a file then the content should be returned as a hash reference with the following keys:

$filename, content => $content, version => $version, # optional, if known }; # content of file stored in the filesystem return { type => file, filename => $filename, path => $path, # full file system path to file version => $version, # optional, if known tmp => $tmp, # optional };

$tmp if set will indicate if the file is temporary or not, and can be used by Alien::Build to save a copy in some cases. The default is true, so Alien::Build assumes the file or directory is temporary if you don’t tell it otherwise.

If the URL points to a directory listing you should return it as either a hash reference containing a list of files:

return { type => list, list => [ # filename: each filename should be just the # filename portion, no path or url. # url: each url should be the complete url # needed to fetch the file. # version: OPTIONAL, may be provided by some fetch or prefer { filename => $filename1, url => $url1, version => $version1 }, { filename => $filename2, url => $url2, version => $version2 }, ] };

or if the listing is in HTML format as a hash reference containing the HTML information:

return { type => html, charset => $charset, # optional base => $base, # the base URL: used for computing relative URLs content => $content, # the HTML content };

or a directory listing (usually produced by ftp servers) as a hash reference:

return { type => dir_listing, base => $base, content => $content, };

sub init { my($self, $meta) = @_; $meta->register_hook( decode => sub { my($build, $res) = @_; … } }

This hook takes a response hash reference from the fetch hook above with a type of html or dir_listing and converts it into a response hash reference of type list. In short it takes an HTML or FTP file listing response from a fetch hook and converts it into a list of filenames and links that can be used by the prefer hook to choose the correct file to download. See fetch for the specification of the input and response hash references.

sub init { my($self, $meta) = @_; $meta->register_hook( prefer => sub { my($build, $res) = @_; return { type => list, list => [sort @{ $res->{list} }], }; } }

This hook sorts candidates from a listing generated from either the fetch or decode hooks. It should return a new list hash reference with the candidates sorted from best to worst. It may also remove candidates that are totally unacceptable.

$meta->register_hook( extract => sub { my($build, $archive) = @_; … });

$meta->register_hook( patch => sub { my($build) = @_; … });

This hook is completely optional. If registered, it will be triggered after extraction and before build. It allows you to apply any patches or make any modifications to the source if they are necessary.

$meta->register_hook( patch_ffi => sub { my($build) = @_; … });

This hook is exactly like the patch hook, except it fires only on an FFI build.

$meta->register_hook( build => sub { my($build) = @_; … });

This does the main build of the alienized project and installs it into the staging area. The current directory is the build root. You need to run whatever tools are necessary for the project, and install them into %{.install.prefix}.

$meta->register_hook( build_ffi => sub { my($build) = @_; … });

This is the same as build, except it fires only on a FFI build.

$meta->register_hook( gather_share => sub { my($build) = @_; … });

This is the same as gather_system, except it fires after a share install.

$meta->register_hook( gather_ffi => sub { my($build) = @_; … });

This is the same as gather_share, except it fires after a share FFI install.

$meta->register_hook( override => sub { my($build) = @_; });

This allows you to alter the override logic. It should return one of share, system, default or . The default implementation is just this:

return $ENV{ALIEN_INSTALL_TYPE} || ;

$meta->register_hook( clean_install => sub { my($build) = @_; });

This hook allows you to remove files from the final install location before the files are installed by the installer layer (examples: Alien::Build::MM, Alien::Build::MB or App::af). This hook is never called by default, and must be enabled via the interface to the installer layer.

This hook SHOULD NOT remove the _alien directory or its content from the install location.

The default implementation removes all the files EXCEPT the _alien directory and its content.