MetaCPAN 2014 QA Hackathon Retrospective

This was my second QA Hackathon, the first being 2012 in Paris. This year, I had a lot of good conversations with various people about MetaCPAN's role in the Perl ecosystem and I pushed a decent amount of code. A lot of it is waiting to be merged, but here's the gist of it.

On the plane on the way over, I did a bunch of cleanup on the API. Mostly style changes, but it makes some of files easier on the eyes. RWSTAUNER was kind enough to fix my failing build after I pushed the code.

When Ovid released TAP::Stream, breaking it out of Test::Class::Moose, it briefly disappeared from MetaCPAN. This reminded me of how annoying it is to find missing modules. So, I added a view to the author pages which will let you paginate through all releases ever uploaded by any author. I think it's quite helpful, even if it's not something that will be used a lot.

I had some good conversations with Paul Johnson about the progress of CPANCover.com. There was an existing pull request to add CPANCover links to MetaCPAN. After a chat with him, he was able to provide a JSON file on CPANCover which MetaCPAN can use to provide the links. This logic is currently in the MetaCPAN front end, but it will likely be moved to the API once the API for CPANCover is stable. On the plane on the way home I decided that the CPANCover code could actually be refactored into its own CPAN module, so I've now done this and pushed the repo to GitHub. I want to run it past Paul Johnson before uploading the module to CPAN.

After speaking with Neil Bowers about an older issue he had opened regarding indexing of 06perms.txt, I added support for that to the MetaCPAN API. Once this endpoint is available, it will give you the owner and co-maint permissions for any CPAN module. I had initially thought that just providing the c, f and m flags that are in 06perms.txt would be sufficient, but Neil set me straight about the nuances of the different flags and the pitfalls that come with making assumptions about them. I used his PAUSE::Permissions module to do this work. My takeaway from our conversation is that, outside of issues with PAUSE itself, nobody really needs to care about m,c and f. What you need to care about his who has ownership and who has co-maint. That's it. Dealing with the flags directly without knowing all of the nuances of how PAUSE works, could take you down a road which leads to heartache.

Moving forward, Neil will likely tweak PAUSE::Permissions to optionally use the MetaCPAN API for permissions lookups, which could greatly speed up lookups of permissions of single modules. So, there's already an immediate application for having this code in the API.

I was also able to add some code samples to the metacpan-examples repository. In the absence of thorough documentation, these code samples serve as a good example for people to get up and running with the API.

At the Paris QA Hackathon 2 years ago, Michael Peters had opened a pull request for guessing the name of a module from .pod files which are missing a =head1 NAME section. I was able to tweak this work and add some tests so that we can merge it into the API. Don't ask me why it took 2 years. I'm not sure there's a compelling reason.

About 9 months ago, Yanick sat down and hacked together some great functionality for adding recommendations to MetaCPAN. The basic idea is that you can anyone can add metadata to modules in the spirit of "SEE ALSO". For instance, you could tag CGI.pm with "see also" Plack. In this way, we could collect metadata about related CPAN modules and, in some cases, which modules have been superseded by others. I didn't wrap up Yanick's work, but I did rebase it to keep his branch up to date with the latest masters. I also now have a better idea of what needs to be done to finish it up. I probably won't be the person to finish this work, but we have extra helping hands with MetaCPAN now. I'll get to that shortly.

I also spent a bit of time with Salve Nilsen, who is working on documenting some things happening in open source. We actually had two interviews. One on Saturday and another on Sunday because the Saturday interview didn't get recorded properly. I'm not exactly sure how that information is going to be used, but he's certainly up to speed on MetaCPAN now. :)

I've had a handful of people contact me over the last few weeks about participating in the Google Summer of Code (GSoC) and the Outreach Program for Women (OPfW). We've asked potential participants to make code contributions to MetaCPAN before they fill out their applications to participate. All of these interested parties are completely new to MetaCPAN, so they require a fair amount of help and supervision. I was able to spend some of my hackathon time on code reviews, evaluations of each student and even a Google Hangout to troubleshoot one student's work. The result of this is that I think we're going to have some strong applications for both GSoC and OPfW this year. The MetaCPAN TODO list is quite massive, so I'm really looking forward to the extra help and I hope that we can get a successful applicant or two.

Aside from this, there were the social aspects of the hackathon, where you get to know people a little better. I can't overstate how helpful this is. Getting to know someone in person can be massively helpful when speaking with them later over IRC etc and it leads to better communication and fewer unfortunate misunderstandings. It's a very good thing.

On a personal note, I had the chance to spend some time with family in Germany on the day before the hackathon and I was able to spend an evening in Paris before my flight back to Toronto. I even worked in a morning run past the Louvre, through the Jardins des Tuileries, along the Seine, down the Champs-Élysées and back to my hotel. I've never gone for a run before boarding a plane, but it was a good way to prepare for a lot sitting down in airplanes, airports etc. I highly recommend it.

I'd like to thank everyone involved with the hackathon, particularly BooK and Laurent, for all of their hard work. There's a lot that goes into organizing something like this and they've done an excellent job. The amount of work I was able to remove from my MetaCPAN TODO list and the amount of knowledge I was able to glean from a room full of smart people have me convinced that my time at the Hackathon was well spent. Getting funded to attend something like this forces you to take it quite seriously, since you're there on someone else's dime. I'm grateful to our sponsors for making this possible. They've done a huge service to the Perl community by providing the funds required to make something like this possible.