(sorry for the lack of links in here. I wrote this on a plane and haven’t had a chance to do anything else to it…)
Overall, it was a very good conference. The organizers did a terrific job and paid great attention to detail. The sponsored evening events were an awesome idea and well-executed. (At least, Friday night’s was, I flew home Saturday evening). It was a relatively small conference at 130 people. I think they can easily have double that number next year, if they want to. But, that would require a change of venue, because the Hotel Palomar’s meeting rooms were filled almost to overflowing. From talking with Chris Williams on Friday evening, it doesn’t sound like they’re interested in changing venues next year, which is a shame because a lot of people will have to miss out on a great conference.
Francisco Tolmasky from 280 North gave the perfect kind of talk to kick things off. I’ve been following Cappucino’s development, so I was not surprised in the least with what I saw. But, Francisco is a polished speaker and many people had not seen Apple’s Interface Builder used to create webapps (via Cappucino’s nib2cib tool). We didn’t get a demo of Atlas, which would certainly have wowed this audience.
I didn’t lump my former SitePen colleague Kris Zyp’s talk in with the other server side JS talks, because Kris was talking a lot more about JS that spans from client to server and using standards such as JSON Path, JSON Query, Persistent JS, and JSON object referencing to move data around seamlessly. Of course, he used Dojo and Persevere as his demo platform, but the ideas he presented can be applied anywhere.
Brian LaRoux’s talk on PhoneGap was quite interesting and entertaining. Brian’s talk had a refreshing lack of gravity, while still providing useful content. For one, he mentioned that Dashcode actually offers good tools for making iPhone web apps. He talked about a variety of iPhone-related JS toolkits, and gave a demo of Nitobi’s iPhone “stimulator” which does a better job of representing how an iPhone app will behave than Apple’s own simulator.
I should note that there were probably 50% fewer bullet points than what I have seen at some other tech conferences I’ve been to. I think the message is sinking in that bullet points suck (except for actual lists of things).
Another former SitePen colleague, Pete Higgins gave a Dojo roundup at the very end of the conference. I saw half of his talk before I had to go to the airport. There are lots of good things afoot in Dojo-land. The new conditional compilation stuff seems useful for a variety of things. For example, Dojo can be built in a super-slim variety (6K) that loads everything dynamically. Or you can dump all of the IE compatibility stuff. With the Bespin project, we have a plan to ship a variety of packagings, and I can see this being useful for that as well.
Pete’s Plugd (which he pronounces “plug-dee” as opposed to “plugged”) project provides a bunch of extra convenient ways to use Dojo and I do hope to see that stuff included in 1.4. Pete says that Plugd will likely add 4K to Dojo’s gzipped size, but I think it’s likely worth it.
Malte did a Track B talk on Joose, his JS object system that is built on ideas from Perl’s Moose. It looks like a very powerful system that provides things like type coercion and traits (called roles here) that you don’t find in the type systems that typically come with JS toolkits. If I recall correctly, Joose weighs in at about 16K gzipped, so it’s not a small package.
Nick Carter gave a Track B talk on his JS ORM, JazzRecord It’s a direct descendant of Rails’ ActiveRecord. It looks like a nice enough package, but seeing his samples made me that much more convinced that Atul is right: SQL does not belong in the browser. The sqlite storage engine may very well, but SQL itself does not. TaffyDB, dojo.data, CouchDB, whatever… just as long as the principal form of expression, persistence and querying is JS. The needs of a typical web client are very different from the kinds of things to which we apply SQL on the server. And, even then, people are starting to realize that SQL is not the best tool for every job.
My talk came immediately after James Duncan spoke about dynamic loading of C code into a SpiderMonkey environment. I lamely brought up ctypes as one approach, but dynamic loading of binaries is not my strength. I suggested that James would get useful feedback for his idea on the ServerJS list, given the number of people who are linking C/C++ libraries up with SpiderMonkey and v8.
I also mentioned that we’re working on binary objects and files and that Jack (the interface) has good prospects given how proven the technique is in Python and Ruby.
I hope that at JSConf 2010, we’ll be able to see some significant apps built on a fairly complete platform.
Conference co-organizer Chris Williams thanked me for my endorsement of the conference, saying that I pushed Mozilla over the edge on sponsoring the conference. I had no idea I had such pull :). Anyhow, Mozilla’s sponsorship apparently had a direct impact on the conference food, which was quite beyond typical conference fare. Thanks to whomever it was at Mozilla who gave the a-ok on this.
As with any conference, the hallway track is among the most important, and I had a good time meeting new people and talking about a range of things. Community-driven conferences do bring in a good collection of people to meet.
I am doubtless leaving people out of this roundup, and I apologize for that. I am sure there are some other JSConf roundups that will provide additional insight. Also, the videos will be showing up online over time, so keep an eye out for that.