It pains me to admit this, but RIBASUSHI was right.
As we rode on the train together into Lyon, he told me that for him
the hackathons are mainly about having discussions, with the hacking
coming later. "Pah!", I said, "I'm here to hack"...
But he's right, the big value of the QA Hackathon is that brings
together 30 or so people who care about Perl, CPAN and the toolchain,
and hothouses them for 3, or in this case 4, days. Some people came
with very specific things to work on, and did so very diligently.
But there were lots of ideas bouncing around, leading to new things,
all generally good for Perl. And so often when problems came up,
people who could help were right there.
I went with some very specific things I wanted to discuss,
and a too-long-list of things to try and get done. I went off-piste
quite a lot, but have come away with a long todo list that I'm fired up
to get working on.
I discussed an abstract
check for CPANTS with Ishigaki-san;
we're going to make that happen.
HAARG and I talked about the Changes format and parsing / module approaches.
I need to write it up, we're both going to write code, then reconvene.
A group of us discussed an approach to handling PAUSE / CPAN email bounces,
and recording whether a CPAN author's email address has gone bad;
I need to write this up, solicit comments, and then try and make it happen,
with various people taking different pieces of the pie.
There were various discussions related to PAUSE and specifically permissions
handling; I sat in on some of these and was very happy to see RJBS,
DAGOLDEN, ANDK and others make them happen.
One of those things was my proposal to change how and when permissions
are checked and assigned.
During the PAUSE permissions work RJBS found two bugs in my PAUSE::Permissions module
relating to the fact that permissions now consider package names case-insensitively.
I released an update with fixes.
I volunteered to try and sort out two lists of quirky cases related to permissions,
one that I generated and one from Rik.
I recently adopted JHI's Graph dist. I spent a bit of time working
on this, but then decided it was just a solo project, so there were better
things to work on at the QAH.
REHSACK asked me to sort out a Changes file for a dist he's involved with,
which is a sideline of mine. In doing this I wrote a script and BOOK showed me
his Git::Repository module.
Two years ago I raised
a MetaCPAN issue
where I ended up suggesting that PAUSE permissions should be pulled in.
I understand them a lot better now.
Olaph and I bounced it around,
and while I continued gadding about, he got started
coding it up. His hacking to chatting ratio was much better than mine
this weekend! Once it's done, and I've learned a bit more about
Search::Elasticsearch, then my PAUSE::Permissions module may
end up talking to MetaCPAN by default.
Olaph, BOOK, BARBIE and I
discussed the CPAN dashboard idea, and whether it belongs in
MetaCPAN. We decided that for the time being, the answer should be "no".
Exactly what it is will be very fluid for a while, so MetaCPAN's not an
appropriate home. Rik separately brain-dumped on me a load of ideas related
to (CPAN) dashboards.
The previous group then hooked up with SJN and SAPER and had a good
discussion about the ways we might use gamification ideas to encourage
people to improve CPAN while having fun. Another thing to write up.
BOOK and I bounced around ideas for how to implement some of these,
and started hacking something up.
During the previous discussions, the idea of 'CPAN Testers latency' came up.
BARBIE pointed out that all the data I needed was available (it always is
when it comes to CPAN Testers!),
and I ended up graphing it.
Walking in Lyon that evening, Barbie and I talked about how these sorts of graphs
might be added to CPAN Testers.
I've raised a number of ideas and issues related to PAUSE. RJBS suggested that
my next step is to submit tests when I think of things. He walked me through a testsuite,
which was impressive. I need to 'level up' on my testing.
I really want to include test coverage stats in the dashboard.
PJCJ was there, and I not only realised that he's the author
of Devel::Cover, but he also runs CPAN Cover.
I had a good chat with Paul about CPAN Cover, and getting data from
that into the dashboard. He's got a great vision for where to take this.
As an aside,
I had too many slightly embarrassing moments when I'd say
"oh, you're the person who does XYZ!".
I had only met 3 of the attendees before this weekend, and many of them
have been meeting up, and working together, for a good few years.
But I was made to feel very welcome, and the only person making me
feel like a n00b was myself.
There are some fearsomely productive people in the Perl community.
Watching people work made me realise I really need to level-up
on git and vim.
Dealing with well-known authors on IRC, via email or in github comments,
it's easy to build up a certain idea of what they're like. It was a bit
daunting going to the hackathon,
but they were all nice friendly people. Normal.
Ok, maybe not RIBASUSHI :-)
In addition to all the 'big ticket' items, a real bonus of the QAH
was that you could say "I wonder what / how / why ...", and where normally you'd
not follow up, here you could answer those questions. Like my wondering what
the time distribution of CPAN Testers reports
looked like.
As was probably clear from the above: I had a great time.
I'm knackered.
Thank you to the sponsors;
I couldn't have gone without their support.
Thanks to DAGOLDEN for initially suggesting I go, BOOK for pushing me to make
it happen, and my partner for letting me leave her home alone with the kids for 4+ days.
And thank you BOOK and ELBEHO for organising it, and Wendy for all
her help in keeping us fed and more.