Beefy Boxes and Bandwidth Generously Provided by pair Networks
Keep It Simple, Stupid
 
PerlMonks  

Automate multi-perl unit testing with Perlbrew/Berrybrew

by stevieb (Canon)
on Feb 09, 2016 at 19:18 UTC ( [id://1154760]=CUFP: print w/replies, xml ) Need Help??

Recently, in Re: Testing in Perl, I said I was working on a script that automates multiple test builds of a module against a number of Perlbrew instances. Below are the brew_build.pl (brew control) script and the test.pl (test runner) script, and here's the git repo.

This works on all platforms I've tested it on (FreeBSD, Linux and Windows). For *nix, you need to have Perlbrew installed. On Windows, Berrybrew is required. You'll also require cpanm from App::cpanminus.

The reasoning behind this creation is due to the fact Travis CI only performs builds on Linux machines, and I wanted an easy way to perform release candidate builds on Windows as well in much the same manner.

It was rather quickly slapped together, but it's simply a prototype. Now that I know it works pretty well, I'm going to turn it into a proper Test module.

In your module, create a build directory in the root, and drop these two files into it. Here are some usage examples:

Run unit tests on all currently installed perl versions:

perl build/brew_build.pl

Remove all currently installed perl instances (except the one you're using), and install three new random versions, and run tests on those pristine instances (short forms for args (eg: -c for --count) are available:

perl build/brew_build.pl --reload 1 --count 3

Install all versions of perl available to Perlbrew, without removing existing instances, and enable verbose output:

perl build/brew_build.pl -d 1 -c -1

Install a specific version of perl, and run tests on all installed versions:

perl build/brew_build.pl -v 5.20.1

Example output (note that if one perlbrew instance fails tests, all processing stops (exit;) and the actual test output for the failed build is displayed along with the perl version so you can further investigate. Otherwise, on success:

% perl build/brew_build.pl perl-5.23.7 perl-5.22.1 perl-5.20.3 perl-5.18.4 perl-5.14.4 perl-5.12.5 perl-5.12.5 :: PASS perl-5.14.4 :: PASS perl-5.18.4 :: PASS perl-5.20.3 :: PASS perl-5.22.1 :: PASS perl-5.23.7 :: PASS

brew_build.pl

#!/usr/bin/perl use warnings; use strict; use Cwd; use Getopt::Long; my ($debug, $count, $reload, $version, $help); GetOptions( "debug=i" => \$debug, "count=i" => \$count, "reload=i" => \$reload, "version=s" => \$version, "help" => \$help, ); if ($help){ print <<EOF; Usage: perl build/brewbuild.pl [options] Options: --debug | -d: Bool, enable verbosity --count | -c: Integer, how many random versions of perl to insta +ll. Send in -1 to install all available versions. --reload | -r: Bool, remove all installed perls (less the current + one) before installation of new ones --verion | -v: String, the number portion of an available perl ve +rsion according to "perlbrew available" Note that only o +ne is allowed at this time --help | -h: print this help message EOF exit; } my $cwd = getcwd(); my $is_win = 0; $is_win = 1 if $^O =~ /Win/; run($count); sub perls_available { my $brew_info = shift; my @perls_available = $is_win ? $brew_info =~ /(\d\.\d+\.\d+_\d+)/g : $brew_info =~ /(perl-\d\.\d+\.\d+)/g; if ($is_win){ for (@perls_available){ s/perl-//; } } return @perls_available; } sub perls_installed { my $brew_info = shift; return $is_win ? $brew_info =~ /(\d\.\d{2}\.\d(?:_\d{2}))(?!=_)\s+\[installed +\]/ig : $brew_info =~ /i.*?(perl-\d\.\d+\.\d+)/g; } sub instance_remove { my @perls_installed = @_; if ($debug) { print "$_\n" for @perls_installed; print "\nremoving previous installs...\n"; } my $remove_cmd = $is_win ? 'berrybrew remove' : 'perlbrew uninstall'; for (@perls_installed){ my $ver = $^V; $ver =~ s/v//; if ($_ =~ /$ver$/){ print "skipping version we're using, $_\n" if $debug; next; } `$remove_cmd $_`; } print "\nremoval of existing perl installs complete...\n" if $debu +g; } sub instance_install { my $count = shift; my @perls_available = @_; my $install_cmd = $is_win ? 'berrybrew install' : 'perlbrew install --notest -j 4'; my @new_installs; if ($version){ $version = $is_win ? $version : "perl-$version"; push @new_installs, $version; } else { if ($count) { while ($count > 0){ push @new_installs, $perls_available[rand @perls_avail +able]; $count--; } } } if (@new_installs){ for (@new_installs){ print "\ninstalling $_...\n"; `$install_cmd $_`; } } else { print "\nusing existing versions only\n" if $debug; } } sub results { my $exec_cmd = $is_win ? "berrybrew exec perl $cwd\\build\\test.pl" : "perlbrew exec perl $cwd/build/test.pl 2>/dev/null"; my $debug_exec_cmd = $is_win ? "berrybrew exec perl $cwd\\build\\test.pl" : "perlbrew exec perl $cwd/build/test.pl"; my $result; print "\n...executing\n" if $debug; if ($is_win){ $result = `$exec_cmd`; } else { if ($debug){ $result = `$debug_exec_cmd`; } else { $result = `$exec_cmd`; } } my @ver_results = split /\n\n\n/, $result; print "\n\n"; for (@ver_results){ my $ver; if (/^([Pp]erl-\d\.\d+\.\d+)/){ $ver = $1; } my $res; if (/Result:\s+(PASS)/){ $res = $1; } else { print $_; exit; } print "$ver :: $res\n"; } } sub run { my $count = shift // 0; my $brew_info = $is_win ? `berrybrew available` : `perlbrew available`; my @perls_available = perls_available($brew_info); $count = scalar @perls_available if $count < 0; my @perls_installed = perls_installed($brew_info); print "$_\n" for @perls_installed; if ($debug){ print "$_ installed\n" for @perls_installed; print "\n"; } my %perl_vers; instance_remove(@perls_installed) if $reload; instance_install($count, @perls_available); results(); }

test.pl

#!/usr/bin/perl use warnings; use strict; use Cwd; my $cwd = getcwd(); if ($^O ne 'MSWin32'){ system "cpanm --installdeps . && make && make test"; } else { system "cpanm --installdeps . && dmake && dmake test"; }

Replies are listed 'Best First'.
Re: Automate multi-perl unit testing with Perlbrew/Berrybrew
by stevieb (Canon) on Feb 10, 2016 at 01:00 UTC

    Input regarding how the module would function, and what it would do is greatly appreciated.

    I'm already thinking commanding remote systems, hooks to various VCS for automation etc.

Log In?
Username:
Password:

What's my password?
Create A New User
Domain Nodelet?
Node Status?
node history
Node Type: CUFP [id://1154760]
Approved by GrandFather
Front-paged by LanX
help
Chatterbox?
and the web crawler heard nothing...

How do I use this?Last hourOther CB clients
Other Users?
Others lurking in the Monastery: (7)
As of 2024-04-19 09:13 GMT
Sections?
Information?
Find Nodes?
Leftovers?
    Voting Booth?

    No recent polls found