Well, here I am in gorgeous Asheville, NC for my third YAPC (Yet Another Perl Conference) North America, taking place June 27-29th. This is by far the nicest of the three conference settings so far (I previously attended YAPC::NA at the University of Houston and at Carnegie Mellon, Pittsburgh).
Asheville is a charming college town nestled in the Blue Ridge Mountains, in the western portion of North Carolina. Looking forward to a little exploring this evening.
But first, Day 1 of the awesome geekfest known as YAPC. My initial impression is how much swankier this is compared to previous YAPCs, but at a registration fee of $100, you’ve got to hand it to the organizers and sponsors. [note: by 'swanky', I do not mean in comparison to the kind of excessive wealth-display conference hosted by many commercial software companies. Different animal.]
Also: I need to get my hands on one of the Grateful Dead / Perl Onion t-shirts I see people wearing. Photo soon, hopefully. [update: they are conference staff shirts. +1 difficulty for obtaining one.]
Okay, here goes…
Day 3 (Wednesday) [most recent postings first]
3:18pm: Well, that’s it for me for YAPC::NA 2011. There are some closing festivities going on, but I’m going to get a jump on driving… heading up to the Erie, PA area to visit my father and family. If you read this, thanks for reading. I also plan to blog OSCON 2011 from Portland, OR in about a month, if you haven’t yet had enough. -Marcus
2:56pm: Hilarious Arduino application: an Arduino-enabled breathalizer which records your blood alcohol count at the time of each of your code commits, allowing you to find your optimal level or drunkenness to code. You can also prevent commits over a certain BAC. It’s dubbed ‘DrinkShield’.
2:22pm: Next talk is about hacking Arduino with Perl. I wish someone had showed me this when I was 10. For that matter, I wish someone had taught me to program seriously at that age. I could play with something like the Arduino all day and night. Immensely cool.
1:55pm: One of the final talks I am attending is about testing with PSGI. There is a module to extend ‘Mech’ called Test::WWW::Mechanize::PSGI.
I am still trying to wrap my head around the (seemingly recent) plethora of choices regarding standalone http servers, middleware and alternate gateway interface in the Perl/webserver space. My basic understanding is that it improves portability/distribution because standard CGI even atop mod_perl/FastCGI ties you to those technologies and/or Apache. May not be a big deal for me since I’m hosting my own apps.
12:03pm: I was delayed in posting because the last session was standing room only, and I found out why. Miyagawa, author of the cpanm module installer (which I love to use) among other things, is just incredible. One of those programmers who gives you the thought, “Wow, okay… I’ll NEVER be that good. EVER.” But you shrug it off and soldier on.
At one point in his presentation (about Plack deployment and many other Perl/webserver issues), spontaneous applause broke out as he demoed an interactive Plack debugger that was ridiculously slick. Guy’s amazingly productive and sharp, not to mention young… the envy, most likely, of more than a few geezers and geezers-in-training like myself in that room. Best talk of the conference.
Miyagawa had conducted a small survey (100 or so responders) of Perl web developers on a number of technology choice questions. When the audience tried to guess the most popular Perl web framework according to the survey results, most of us muttered ‘Catalyst’… but what did the next slide reveal? The most popular web framework was ‘in-house’. Had to laugh, had to nod. My own homegrown, hand-rolled framework, had I been a survey responder, would have added to those results, for better or worse. The Perl community has a very strong streak of DIY.
10:34am: The whole point of Parrot is to make it ‘dead simple’ to create new languages. There’s a bit more to it than that I think, but that’s not a bad soundbite. Computer languages suffer from the same thing that natural languages do: Parrot is really about tackling that problem.
10:28am: Wow, very cool idea that Parrot got from Moose: storing deprecation data in structured data files, instead of just mentioning it in the change log. A great way to handle deprecations for any API. I’ll have to remember this technique and have a look at the Parrot repo if I ever want to steal their format. They even define regular expressions which would help you find deprecated calls in your code.
10:01am: First up today for me, is Jonathan Leto with a talk entitled ‘A Visual Introduction To Parrot Virtual Machine’. Calls PIR “the most pleasant assembly language you’ll ever see”. Stands for ‘Parrot Intermediate Representation’.
As expected, this Parrot talk is over my head to a certain extent, although I have gotten a lot better at following the basic gist of these talks. For instance, I understand his point about communications between PIR and C code limiting the opportunity for many optimizations, but I don’t really understand it down in the details. But I don’t need to. I don’t think I’ll be writing any virtual machines any time soon, or implementing my own new language.
Day 2 (Tuesday) [most recent postings first]
4:26pm: Calling it a day at this point. This conference is either getting better, or I am better able to follow the various talking points than I was when I attended my first, four years ago. Maybe both. See you tomorrow.
4:05pm: By the same token, as these one-liners grow in complexity, one realizes why they end up in scripts. Geesh. Readability == quite important.
3:55pm: Actually learning a lot about one-liners in this session, excellent presentation. -n for implicit looping is awesome. I am lazy (or not?) for never having learned that.
3:29pm: I’m going to round out the day with Walter Mankowski’s talk “Essential Perl One-Liners’. I’ve actually never been much of a one-liner type guy, so maybe this will inspire me. Walter came down from Drexler University in Philly.
On a random note, one thing I am hearing a lot here, aside from recruiting, is developers on cellphones helping younger or newer developers back at their companies. Apparently this kind of project mentoring and handing-off is an extremely common part of the workflow at many orgs. This has been a part of my working life lately, too.
3:18pm: I enjoyed that talk. Obviously a very talented developer. It was also interesting to see how he’s been pondering the next version of WebGUI (the parent company of The Game Crafter) using Modern Perl techniques. WebGUI is itself a CMS but it will be sitting atop lower-level frameworks. The whole trick as a developer, I still think, is picking the right ‘level’ to be working at for your needs. I think most IT people work one or two levels above where they ought to work, depending on others too much and feeling the pain when dependencies become letdowns (most especially in the commercial space).
3:03pm: Now talking about ElasticSearch but also mentioned Sphinx which I’m a bit more interested in. Not entirely sure I have internal search needs of that magnitude but it might be worth trying out at some point. ‘Search fuzziness’ is certainly not something I’m going to get out of SQL ‘LIKE’ statements, for instance.
2:55pm: Oh! Cool module tip… Imager. Think of it as Image::Magick lite, and for anyone who has ever tried to install Image::Magick, you might appreciate the ‘lite’. Apparently does basic image alterations like thumbnail creation quite well and simply. Must remember this one.
2:47pm: Speaker claims that Image::Magick can do anything that Photoshop can do, but in Perl. Sounds fun. Reminds me a lot of working with PDF layouts [shudder] in Perl, which we’re hoping to avoid completely in the future.
Ha, look at that. Image::Magick can output pdf format.
2:40pm: Switching to discussion of Email::Sender::Simple. A lot of the advantages are about portability, which doesn’t concern me too much, since most of my software is in a hosted environment at this time. Sticking with MIME::Lite for now.
2:34pm: Speaker highly recommends Email::Mime::Kit and Email::Sender::Simple. Claims the first is good at formulating spam-filter-proof emails, and sending attachments properly. Email::Mime::Kit is an email templating system as well… which is nice, but an investment. The templates are ugly though, whether for plain text or html emails. Not digging that too much.
2:21pm: Next up, “The Game Crafter: A Perl Success Story”. JT Smith, the speaker, is also an organizer of next year’s YAPC which will be in Madison, Wisconsin.
Fascinating, “The Game Crafter” is a service for prototyping board and card games, written in Perl. It’s a web-to-print service for games, similar to how Lulu is web-to-print for books.
2:16pm: Hmm, the frontend of this is called wysihtml5 and it’s on github. Maybe worth checking that piece out…
2:09pm: In a nutshell, the guy is sanitizing the HTML by parsing and then rebuilding it, to support HTML cut-and-pastes, but discriminating against unwanted tags (such as ‘script’) etc. I’ve seen this done poorly a thousand times with WYSIWYG editors, where they mangle your HTML beyond usefulness… but this seems fairly clean. In general though, I’d prefer not to solve this kind of problem myself, relying instead on a really nice JS library. I’ve used TinyMCE in the past but I’m looking to switch to something jQuery.
1:51pm: Next up, a presentation on HTML5::Sanitizer, which apparently hasn’t been published to CPAN. Speaker Uwe Voelker is talking about a project to embed a smartish WYSIWYG editor into the XING website. XING is like LinkedIn, but more popular in Europe.
A key piece of this work is XML::LibXML as the parsing tool. Ah… he is replacing special characters with HTML entities, one by one, feeling that XML::libXML doesn’t provide all the best escapers. Been there, felt that pain. I still marvel that we haven’t moved past this kind of thing, but that’s code for ya.
11:53am: Jesse’s best quote: “There are certain gift-horses I would like to inspect the teeth of”…. in reference to people with Perl feature ideas they aren’t prepared to work to implement.
11:34am: Ok, another endorsement for autodie. They want to bake autodie’s improved error handling right into how Perl itself could better handle this stuff, in the way subs return on failure, etc. So, clearly, using autodie today in Perl 5 is the way to go… but stay tuned.
11:31am: Something I did not know: in Perl 5.12, ‘use v5.12;’ includes ‘use strict;’. This shows how Perl 5 is slowly moving towards similar saner defaults as we are also expecting in Perl 6. Neat.
11:20am: Jesse is proposing a new way to declare not just the oldest Perl version you want to use, but the exact version, when writing ‘use v5.16′. This is not real yet. But it would allow Perl to move forward more easily while maintaining backward compatibility.
10:56am: Next up, a talk about ‘Perl 5.16 And Beyond’ by Jesse Vincent, the current Perl 5 pumpking. Session is packed, Larry is here, my coffee refill is STRONG. No more making coffee in the room. That’s whack, this is the stuff.
10:41am: Turns out to be a talk about this module which you could add to your OO toolset. An ‘announcement’ is a particular kind of object design pattern focused on events that other objects may want to know about. The idea started in the Smalltalk community in work done by Vassili Bykov. Definitely interesting stuff, although nothing I can immediately apply in my procedural designs.
10:31am: Now for Shawn Moore, speaking on ‘Announcing Announcements’. No idea what to expect here.
10:21am: I’m pretty certain that I’m not looking forward to working heavily with data caching. At present, all my apps access data right off the disk on every query. This clearly doesn’t scale in certain ways for certain applications and it does make me wonder when and where I’ll hit a ‘ceiling’. The kinds of problems programmers have to deal with around data caching are ugly.
10:14am: This guy is melting my brain. He’s basically talking about code optimizations that prevent the access of data until you really need it… to defeat certain caching inefficiencies… but I felt my brain leaking out my left ear sometime around his description of ‘trampoline subroutines’.
9:59am: The first session for me today is Steven Lembark’s ‘Virtuous Laziness For Your Data’ subtitled ‘Doing It Once And Knowing You’ve Done It’.
Day 1 (Monday) [most recent postings first]
5:07pm: I’m calling it a day. Good first day, I thought… nice mix of technical and less-technical stuff. Time to go see what Asheville is all about!
4:08pm: Now for a likely lower-tech session called ‘Telecommuting’, subtitled ‘…Or How To Survive Work Without a Water Cooler.’ Telecommuting has always been a pet topic of mine, since I can’t think of too many other reasons to seek a career in computer programming. It certainly isn’t for time outdoors or the dating opportunities.
[More rumblings about everybody's company hiring at this event. Perl economy == not so shabby.]
Presenter works 100% of the time out of the office… there is only one person left full-time in their office. He is in Manhattan…. Manhattan, Kansas.
One good piece of advice: don’t try to telecommute if you don’t love your job. He didn’t really explain why, but it’s obvious… if you’re not self-motivated by the joy of coding, you’ll be even less motivated when there’s no one else physically present. If you love it though, the inverse may be true as you find more quiet and space to concentrate.
Another salient point: too much security is a problem. Security vs. convenience once again. Typical complaints… which I share… about VPNs.
3:29pm: Matt Nash now presents his talk entitled ‘Perl Helped Me Graduate’. Matt is a UNC Computer Science grad who used Perl for his senior project. The projects were doled out as part of a corporate partners program and this was an actual industry project for AT&T.
Cute how he bargained with his (younger… was was a non-trad) classmates to let him do the backend in Perl as long as they could do the front-end in Adobe Flex… because they thought it would help get them jobs. Really? Amazed how people always look at technology as ‘what’s hot this year?’ rather than the more advisable ‘what’s truly got legs?’… especially when they’re developers. Who wants to rewrite in two years?
2:56pm: Sometimes it’s easier to just stay in the same room, especially when the next talk is entitled ‘The Art of Klingon Programming’. Paul Fenwick is in full Star Trek gear, but his laptop is misbehaving right now.
Okay, he tamed the laptop. This talk is brilliant and strange. He backs into the idea of not doing things implicitly in Perl by way of some hilarious ‘translation failure’ jokes… in a Klingon nightclub scenario. Yes, you’d have to be here. Really missing Bill Costa and Paul Sand right now.
This turns out to be a talk on autodie, which I have not hitherto appreciated. And of course here I am at YAPC so I am hearing about it from the author himself. Should I be using this? I think maybe I should. Wondering if Henninger would be interested in this… it’s lexically scoped, and excellent for error handling on system commands especially.
This gets rid (and more) of your ‘or die’ statements. Awesome. Come standard with Perl 5.10.1 and can be installed on 5.8+. I am curious how this might work with Costa’s version of FatalsToEmail.pm. Since Costa’s module mostly rewires die(), maybe it just sits down on the chain after autodie. That’d be great.
Excellent presentation. I like to have at least one take-home worth implementing from every conference I attend, and autodie might just be the thing.
2:25pm: Mark Fowler’s talk titled Pimp Your Mac With Perl. If you hit that link, you’ve just been schooled on how to give a talk at a conference and promote it well. Think of a fun talk title, get the domain, publish your talk and some related links.
He’s going into some detail on using launchd on OS X in lieu of init.d for those familiar with Linux. Good talk.
12:37pm: Lightning talks end.
12:34pm: Next talk is promotional for the Pittsburgh Perl Workshop. Interesting impromptu vote, show of hands for people who consider themselves primarily sys admins vs. primarily app developers. Waaaaaaaaaay more people who consider themselves app developers in the room.
12:23pm: more recruiting… NYC… Amsterdam… telecommute…
12:28pm: Another quick talk, this on perldoc. Usually only works if installed, so he has written cpandoc, which does a fetch on anything in core Perl or CPAN. Very nice for people who prefer not to leave the command line to read docs.
One thing that strikes me: using cookies is a great convenience for users, but holy cow does it open up security concerns. At that point you should force all DB-altering calls to be POSTs, not GETs for instance. Ought to be doing this anyhow for other reasons too.
12:16pm: Next: a talking about getting good debug on his dynamically generated SQL. Looks like he wrote some custom code to do this… I think DBI::Profile does just fine for this but maybe I am missing for something.
12:09pm: Next: another threading library for Perl. Ah, threading. I have occasionally messed around with forking background processes for long-running tasks, but I sincerely hope to never have to worry about threads again. I enjoy low-level work but I’d love to see process/thread management become as uncommon as memory management in HLLs.
12:02pm: Next up: ‘Slicing and Dicing Plant Genomes with Perl’. I’d secretly (whoops!) rather be working on stuff like this than on business apps… but I might be a bit behind in my knowledge of genetics…
This guy studies everything in the Nightshade family. Tomatoes, potatoes, petunias (petunias?). I think I need to bone up on my botany, too.
Interesting to know: there are a lot of opportunities to improve the applications and interfaces available to scientists.
11:57am: Now we are hearing about an effort at benchmarking template modules. Shoot, he isn’t including HTML::Template in the comparisons even though he supports benchmarking it. Oh well… it’s fast enough for me.
11:52am: Next lighting talk is a primer on creating a basic CRUD (create/read/update/delete) app with an object model. Apparently a Mojolicious framework plugin that sort of seems like a combination of an object modeler/dispatcher that also scaffolds your UI a bit.
11:51am: Hard not to notice the recruiting going on here. More than in the past it seems. Is the future brighter than ever for Perl devs?
11:46am: Next lightning-talk guy is speaking about version-controlling your home ($HOME) directory in the UNIX environment. Actually this seems more than a management script for multiple working directories, handy for coders on multiple projects…
11:40am: First lightning talk: a guy is pitching the Stackato cloud platform from Active State. Supports Perl, Python and Tcl. Supports any PSGI-compatible framework. (Is this a basic requirement for frameworks nowadays? I still don’t entirely understand the problem space that PSGI attacks)
11:16am: Keynote has concluded, looks like lightning talks next in the big room here. Not sure I recall these being a main feature before but it’s a good idea; lots of information, very quickly. We’ll see how well I blog it, though.
11:01am: Larry gets a few hugs (somewhat of a lark to cure his ‘depression’) from the crowd, and some thank-yous (not a lark) for creating Perl and allowing all of us to have careers.
10:58am: I am struck by how Larry focuses on very human issues when it comes to fostering community and collaboration… not tools. We all use collaborative tools but the real problems with cooperation are within us. Obvious, but shouldn’t we talk about it more? Try to think about solving it more? Seems like we talk about tools 90% of the time and tools won’t save the day.
10:47am: Interesting observations on ‘hares vs. tortoises’ and ‘merchants vs. craftsmen’ to describe different work styles and how they are each valuable. Also early vs. late adopters, detail vs. big picture people, leaders vs. followers…
10:42am: Larry’s wife walks in.
10:36am: Larry is typically philosophical. He’s talking about community now, ‘hugging trolls’, using social martial arts and keeping core Perl contributors sane. Imagine piloting an all-volunteer project composed of hundreds of geniuses and eccentrics. No wonder Larry thinks about the psychology of his community a lot.
10:19am: Larry reminds us that he is our BDFL. Benevolent Dictator For Life. It’s true, and we’re glad.
10:10am: Larry is about to come up for the keynote.
This is reposted from the report posted on software.unh.edu.