All the Perl that's Practical to Extract and Report

CPAN::WWW::Testers

acme on 2003-08-01T09:29:23

Yesterday evening, after a total of six hours of hacking, I released CPAN::WWW::Testers, which downloads and presents CPAN Testers data. From the docs:

CPAN Testers is really a mailing list with a web interface, testers.cpan.org. testers.cpan.org was painfully slow. I happened to be doing metadata stuff for Module::CPANTS. This is the result. It's alpha code, but using it anyone can host their CPAN Testers website.

Actually, I've just checked it this morning and testers.cpan.org is still horribly slow. Getting all the distributions which begin with 'A' takes 50 seconds for testers.cpan.org and 0.4 seconds for testers.astray.com. This is because I generate static files.

Actually implementing it was fun. The turning point where everything suddenly started being clear was realising that testers.cpan.org was really an interface to the NNTP archive of the cpan-testers mailing list. The rest was done using Net::NNTP, Email::Simple, caching, DBD::SQLite, Sort::Versions and the Template::Toolkit.

It obviously needs a little design work and some computer-readable way of reading results. Anyone got some spare time?

Phew, that's secret project number one down...

Freaky coincidence

koschei on 2003-08-01T10:07:34

I was also fed up with testers.cpan.org, so I did dellah.org/testers.

There's one program that is cronned to fetch update via nntp and stuff them in an SQLite. However I still present them dynamically via Mason and a badly configured mod_perl. And it's still faster than testers.cpan.org. I do point people to nntp.perl.org rather than mirror the bodies of all the tests.

Mine was deliberately made to look much like testers.cpan =)

As a side effect: a picture of uploads per day.

Re:Freaky coincidence

hfb on 2003-08-01T10:18:27

WHERE HAVE YOU BEEN? :) If you'd like to take Acme's module and give it that interface I have a Sparc in London you can host it on :) pretty please? :)

Ordering bug

IlyaM on 2003-08-04T15:38:14

This doesn't seem correct to me (2.12_xx comes before 2.101).

Re:Ordering bug

IlyaM on 2003-08-04T15:40:49

Also there is something wrong with number of passed and failed tests:
2.12_02 (5 FAILs, 4 PASSes)
96696 PASS sun4-solaris
96761 PASS MSWin32-x86-multi-thread

2.12_01 (6 FAILs, 5 PASSes)
95765 PASS sun4-solaris
96190 FAIL MSWin32-x86-multi-thread

Re:Ordering bug

acme on 2003-08-04T23:34:02

I'm using Sort::Versions in reverse mode, so most recent comes up top. According to Sort::Versions (and it seems right to me too), 2.101 is after 2.12.

Re:Ordering bug

IlyaM on 2003-08-05T07:53:30

It is wrong. Version numbers of Perl modules are compared as floats. So 2.12 should come after 2.101.

Re:Ordering bug

acme on 2003-08-05T09:46:57

[sound of leon rustling around in perlmod, then perlmodstyle]

Oh, you're right. Fixed, thanks.