Recently in Perl Category

Accepted Grants for 2010Q3 doesn't make a clear border line and I got 2x congratulations that my grant was accepted. The truth is only the first three each quarter are. I just "scored" at the last 5th, non-rejected place.

When I showed Benjamin (a college of mine) my TPF 2010Q3 grant proposal, we slipped to a discussion about deploying Perl software. Nice (and recent) list of different approaches can be found @modernperlbooks.com. To sum it up => TIMTOWTDI. Which is good, but none of those is perfect. The Perl+CPAN world is way too complex.

During out discussion with Benjamin I proposed an idea of shipping the application altogether with the OS. Insane? The base Debian system is ~190MB, all the rest is needed for the application. Then deployment will be a matter of running this system on a virtual machine, somewhere in the cloud or in a simple chroot. (btw any Linux distribution can have any other Linux distribution working in a chroot) The files will never clash, all the "machines" would be dedicated. No fear of putting files where they belong to.

Xacobeo a Perl GUI to visualize XML and perform XPath queries just entered Debian unstable - http://packages.debian.org/sid/xacobeo.

It is amazing what happened to the Python thanks to just one company. The killer feature of that company is that they can handle and manage millions of servers and even give them for a public usage => "..., there are no servers to maintain: You just upload your application, and it's ready to serve your users.". I would like this to be true one day for Perl too!

I really do. In coming months (years?) I'll try to plug together « MiniCPANInject + SmokeTesting + ContinuousIntegration + Virtualization + Debian + Monitoring » to create an architecture for Debian-Perl hosting. Actually who cares about OS used, when the uploaded code just works, right?... A hosting that is aware that one server is not enough for everyone and will offer an architecture for development, user-acceptance, staging and production environments to start with a project from $day == 1.

I wanted to get some CPAN authors sample. Not everyone as there are too much (> 8000). So I've created this two on-liners to get the CPAN authors whose module is packaged for Debian:

apt-cache search perl | perl -lane 'next if $_ !~ m/^lib(.+?)-perl\s/; $m=$1; $m=~s/-/::/g; print $m;' > /tmp/deb-perl.txt

zcat ~/.cpan/sources/modules/02packages.details.txt.gz | perl -MIO::Any -lae 'my %x; BEGIN { %x=map { $_ => 1 } split "\n", IO::Any->slurp("/tmp/deb-perl.txt") }; while (<>) { @F=split /\s+/; $m = lc $F[0]; if ($x{$m}) { (undef,undef,$a)=split "/", $F[2]; print $a; }; }' | sort | uniq | xargs echo

And here is the result (842 out of 8270 total):

Joel shows that StackOverflow is (||was?) running on 2 servers (+1 backup) using C# and SQL Server. Impressive :-) seek to 24:56 to see the full performance+tech table.

There is also a Perl+Perl6 mentioned in this talk, seek to 38:00 to hear the #1 thing you should never do...

is just a base module responsible for loading extension (Data::Keys::E::*) during the object build time. It just makes sure one of the extensions supplied get() and set() method. The rest is up to the extensions.

get() should be called with one argument a $key and return value for it. set() should be called with $key and $value arguments returning modified (if it was) $key.

So far I've created this extensions:

• Store::Dir - folder storage
• Store::Mem - in memory storage
• Dir::Auto - auto create folder when needed
• Key::Adapt - change key with a callback
• Key::Auto - auto create key via callback
• UniqSet - a key can be set only once
• Value::InfDef - inflate/deflate values
• Locking - get/set locking
• Dir::Lock - uses additional folder to lock files
• Dir::LockInPlace - place locks directly on the stored files

How I want to use it now? Value::InfDef to help reading folders full of JSON files. Key::Auto to store data as Git does - based on the hash of data. Key::Adapt to easily access Debian distribution files in the pool folder or CPAN distributions files based on the AUTHORID/DISTRIBUTION in CPAN mirror folder structure.

For now there is just folder storage as I want to use it for some tasks, but it should be quite easy to create any key/value storage extension.

So the Perl QA Hackathon 2010 is over. My part in it was mainly to help with organizing, but with a specification from Gregoa and a help from Potyl a new debian-perl quality script was created - patchedit.

Patchedit works in two modes edit and check. In the edit mode it will fire-up an editor and optionally add a missing patch description fields. In the check mode it can test the fields and output TAP based on the ok/fail. Thanks to this and TAP::Formatter::HTML it was easy to create a page with the patch "health" status. Here is how:

find -name series | perl -MIO::Any -MFile::Basename=dirname -ne 'chomp; my $s = $_; my $bd=dirname($s); my @pl=map { $bd."/".$_ } grep { $_ !~ m/^\s*$/ } map { chomp; $_ } split("\n", IO::Any->slurp($s)); print map { $_."\n" } @pl;' | xargs prove --exec "patchedit check -o" -m -Q --formatter=TAP::Formatter::HTML > output.html

The output page can be found here.

You can watch the final stand-up video with the achievements at brian d foy blog.

I didn't know about this easy way to locate Perl modules in Debian packages.

$ dh-make-perl --locate Foo::Bar
Using cached Contents from Fri Mar 26 10:25:51 2010
Foo::Bar is not found in any Debian package

$ dh-make-perl --locate Moose::Meta::Class
Using cached Contents from Fri Mar 26 10:25:51 2010
Moose::Meta::Class is in libmoose-perl package

$ dh-make-perl --locate Test::More
Using cached Contents from Fri Mar 26 10:25:51 2010
Test::More is in Perl core (package perl) since 5.6.2

So there is a ready made solution how to locate Perl modules in Debian packages. The '--locate' option is in Squeeze (testing) version of dh-make-perl or in the source repositiory.

From http://dev.catalystframework.org/svnweb/bast/revision?rev=8823

use Test::Deep qw(eq_deeply);
...
return eq_deeply($bind_ref1, $bind_ref2);

vs

use Data::Dumper;
...
local $Data::Dumper::Useqq = 1;
local $Data::Dumper::Sortkeys = 1;
return Dumper($bind_ref1) eq Dumper($bind_ref2);