Maintaining possibly unused modules
My participation to the CPAN Once a Week contest forces me to find a module to create or update every week. And because I don't want to cheat, it has to be a meaningful change (I also try not to make a new Acme::MetaSyntactic release every week).
This week, I decided to look at my module list through the filter of CPAN Testers. One of them was really not looking good
Given that it's a web scraper, it's not a surprise that the test suite started failing at some point.
The excellent CPAN Testers Analysis site made it quite easy to actually pinpoint that it was all a date issue and even to find out exactly when it started failing:
12011910 FAIL 2011-05-06T14:46:43 # all FAIL from that point on
11941549 PASS 2011-05-03T00:25:17
Anyway, it turned out relatively few things had changed, and I could fix the module in a few hours of work.
There was a tiny bit of discussion on questhub, which led me to write this post. Here's the message we want to get through:
When you find a module is broken, do not just move on, and please let the author/maintainer know. They might have ignored the FAIL reports, or think their module is not really used. After all, it's easier to ignore a machine-generated report than a personal email (most web forms for reporting bug end up sending us emails, anyway). Letting them know they have potential users might give them the motivation to fix the issue.
The last person to contact me about WWW:::Gazetteer::HeavensAbove did so in 2009.
I wrote the module in 2002, when I was still struggling to actually create stuff worthy of being put on CPAN. This one is just a successful script that I used to dump the http://www.heavens-above.com/ database for a friend. It had an interesting enough "paging" algorithm, that I wanted to put it on CPAN and talk about it at YAPC. So when I found out about WWW::Gazetteer, I had a good excuse to actually release it. I was actually surprised to have a genuine user in 2009.
I still like the way it gets past the 200 results limit of the web site, but I also think it's not a very useful module. At least I scored another week in the contest, and CPAN is somewhat less broken for it.
Pink.
Thanks for writing this.
You say people should send you personal e-mails with bug reports as that will trigger faster response, and as a proof you tell about an e-mail you got in 2009, to which you quickly responded 4 years later....
I always knew you were laid-back, but I am impressed.
grin x 10
The 2009 bug was fixed in 6 days according to RT.
Actually this year's work was triggered by the CPAN Once a Week contest and the sorry state of my module test suite. The 2009 bug, as Toby remarked, was fixed within a week.
I do have 4 years old opened bugs... but not for my unused modules! :-)
Probably I should have added to my comment "I know these are two different issues, I just wanted to make a joke. You don't need to point to RT to prove me wrong."
It also turns out that some people never see failing cpantesters reports, so bug tickets from genuine people are the first indication they have that something might be wrong.