"Patience is a virtue", but not for some authors it would seem. Up until last week the Reports and Statistics sites were updated once a day, as the databases behind them were also only updated once a day. As of Monday 26th January, the updates now occur twice a day and will hopefully occur more frequently in the future, although at the moment we're restricted by the amount of time it takes for the Reports site to update (currently 7 hours). As a consequence the last report entered into the database can be from several hours earlier, as it takes a while for the NNTP server to catchup with the mail posts. The old site runs continuously, and thus it often picks up more recent reports sooner than the new site. However, the old site has several bugs that have been fixed in the new site, so there is a trade off. Having said that, please don't bug the current maintainers of the CPAN Testers ecosystem about the old site, as aside from the fact we don't have access to the servers, the old site will be depreciated completely once we move to CPAN Testers 2.0. with a dynamic Reports site.
At the beginning of January the cpan-testers mailing list received it's 3 millionth post. Considering the 1st post was in August 1999, you might be forgiven that that is a fair number to cover just over 9 years of CPAN Testers. However, it's worth bearing in mind that the 2 millionth post was received in August 2008, and perhaps more significantly the 1 millionth post was in February 2008! That's over 1 million posts in the last 5 months, and over 2 million in the last 11 months. The growth rate of testing has been staggering in the last year, and with more people considering other platforms, I expect it to increase further this year. However, it's easy to spot the number of posts, as that is just the NNTP ID, but there are a number of posts that are conversations, questions, invalid reports and PAUSE upload posts, so to confirm the current status I set about writing a short program to help me calculate it all. The code has now been integrated into CPAN-Testers-WWW-Statistics and the results can be seen on the Interesting Stats page. As we hit the 3 million post this month, I was expecting the 3 millionth accepted report hit during February, but I should have known better :) As can now be seen from the tables, Justin Reynolds was the lucky soul to post that particular report on 21st January. The tables make interesting reading, with Chris Williams (unsurprisingly) doing his best to hit every notable entry in the database :) Incidentally, the new tables now also indicate the last report processed, so you can now know how recent the updates are.
Also at the beginning of January, David Westbrook offered to help improve the test suite for CPAN-WWW-Testers, which was very gratefully received. The old test suite was in need of attention, and David has managed to bring it dragging and screaming up to a much more respectable level. To highlight this, David sent me a before and after Devel::Cover comparision of his changes:
File stmt bran cond sub pod time total -------- ------ ------ ------ ------ ------ ------ ------ BEFORE 13.1 2.6 0.0 49.0 100.0 100.0 12.8 AFTER 92.5 58.0 41.3 93.9 100.0 100.0 83.3
With the new tests as a base, a fresh TODO list and some free time, the codebase has been improved even further. Adding even more tests have pushed the coverage rate to over 90%, so many thanks to David for his efforts, it really is much appreciated.
A further move towards the dynamic Reports website, has seen the introduction of a new web service. It is primarily for developers of sites that use the cpanstats data, but find the burden of grabbing the current SQLite database, or parsing YAML files a nuisance. You can now access portions of the data via an AJAX/CGI request. If you wish to find out more details, please let me know and I'll furnish you with how to use the service.
The 2009 QA Hackathon event in Birmingham is drawing closer and will be featuring work on the CPAN Testers infrastructure. With several key collaborators planning to be there, we're hoping that a lot of the plans to get CT2.0 up and working consistently for all smokers will come to fruition.
We topped 137 testers submitting reports last month, so thank you very much to everyone involved. The mappings this month included 36 total addresses mapped, of which 18 were for newly identified testers. This actually prompted me to think about how many new testers we actually get each month. Not only that but how many testers drop off each month too. So I've now created another graph on the Graphs page, that highlights the trends. Interestingly it would appear that when we have a large number of new testers, they often only last a month. Thankfully, the general trend is good.
