We needed to upgrade from EL6 to EL7 at work. After the upgrade, we noticed some of the JSONs returned from the APIs had changed: some numbers were suddenly enclosed in double quotes, while some others lost the quotes, which broke customers’ code written in strongly typed languages (e.g. Java).

We used the system default perl to drive our web application, using libraries provided by the vendor. While Perl’s version changed from 5.10.1 to 5.16.3, the version of JSON::XS responsible for creation of JSON data jumped from 2.27 to 3.01. Both the changes contributed to the problem.

CPAN Pull Request Challenge selected App::GitHooks for me in February. As usually, I started by browsing the issues, and my attention was drawn to Tim Bunce’s request to add a way to skip given hooks.

Two Blocks with the Same Name

In January this year, I got Jemplate in the CPAN Pull Request Challenge. The module implements Template::Toolkit in JavaScript, so you can transfer the burden to process the templates from the server to the client.

One of the open issues in its GitHub repository caught my attention: Jemplate compiled all the templates coming from different files into one large JavaScript code-block. If you declared two blocks with the same name in two different files, they’d end up in the same hash in JavaScript, and only one would survive—but you couldn’t tell which one. Keeping block names unique across files probably isn’t part of common practice, so making the module warn you in such a case sounded like a reasonable requirement.

The issue

In the last month of 2016, I was assigned OrePAN2 in the CPAN Pull Request Challenge. When browsing its issues on GitHub, I discovered #47:

The Report

In a GitHub issue discussion by my former colleagues, I noticed weird errors being reported. I was curious about the cause, but not enough to dig deeper, and as nothing happened, I forgot about it.

But then they asked me for a favour, and I agreed to help them with a project. But, to my surprise, I wasn’t able to install the tools I originally helped to create, getting the same error as the poor GitHub guy. Quick check of the CPAN Testers showed problems in Treex::PML, one of the underlying libraries, all the way down to 5.20.0. A typical report would look like this: