robjsoftware.info

(Warning: relatively low technical content ahead!)

Rats, missed another blog-posting week here. A week ago I finally got my GWT conference presentation done. I was rather nervous about whether I was hitting the right technical level, so I mailed it to a few GWT team members, and Bruce Johnson (the team lead) wrote back and said, basically, “It’s perfect.” Thanks, Bruce! Eased my mind tremendously. The conference is coming up next week.

Then on Tuesday, Mass Effect shipped, and I was instantly lost in space. I’ve gone back and forth on whether to talk about gaming here, but this title pushed me over the edge. I’ve been a computer gamer ever since I’ve been a computer user — my very first experience with a computer was playing Spacewar on a PLATO timesharing terminal at my friend’s father’s college. And gaming has become a true lifelong hobby. Mass Effect is a great example of why that is.

Programmers inherently love systems. A truly intrinsically motivated programmer will spend hours thinking about how something fits together, and more hours tinkering with it and seeing what happens. A great game also has a systemic internal structure that rewards tinkering and exploration. So it’s natural that so many programmers are gamers.

Moreover, many programmers are thrilled by advances in technology. The more powerful our computers become, the more we can do with them. And computer games are definitely the most widely visible hardware-pushing applications. Just take a look at some of these screenshots. A $280 XBox 360 can now do real-time facial animation that looks a lot better than ANYTHING being done five years ago (real-time or pre-rendered). And it qualitatively changes the nature of the game, when your character and the zany aliens they’re talking to both have facial emotions and lip-syncing that are so much closer to realistic that you can almost lose yourself in the illusion. I’m not claiming that Bioware’s crossed the uncanny valley just yet, but they’ve definitely taken huge strides in that direction, and they’ve hooked me but good.

I expect to be a gamer for the rest of my life, since games just keep getting cooler and cooler as hardware gets more and more powerful and we learn more about what to do with it. It’s a perfect hobby for a systems-oriented, game-addicted technophile.

So Mass Effect took over my after-the-family’s-asleep life, and stole my blogging cycles last week. Thanksgiving was also pretty hectic — we cooked an 18-pound turkey for our church get-together, which took me all day while my wife juggled our three-month-old and our (almost-)three-year-old. That left us pretty wrecked and we laid low the rest of the weekend.

Meanwhile, on the hacking front, I have continued to feel blocked by the whole Seam 2 broke my code issue. Actually, technically, it’s that Seam 2 broke the G4JSF code. That code was originally written by the Ajax4JSF team, but they have abandoned it. So the question for me is, how motivated am I to fix it? (And, potentially, to continue fixing it as Seam, JSF, and GWT continue to evolve?)

I originally started on this whole Seam / GWT / JSF project because I had ambitions to create a peer-to-peer distributed blogging system. I made a lot of progress on that system (at least as far as creating a Seam-based model for representing version trees with Java persistence, a la Mercurial). Then I wanted to start on the UI. Which led me to GWT, and the realization that the existing G4JSF library didn’t couple gracefully with GWT RPC. So I patched GWT, and then G4JSF, and that led me to presenting to the GWT team at JavaOne last year, which then got me the opportunity to speak at the GWT conference next week. All of that was quite unexpected and quite appreciated — particularly the speaking opportunity!

BUT, going back to my original project, the peer-to-peer distributed object system lost its mojo somewhere over the last year. It led me down the GWT / Seam / JSF path, which was really interesting and connected me to the open source world like never before… but somehow the journey became the destination, and the original goal no longer feels as interesting. Frankly, I’m not feeling very enthusiastic about being the sole owner of the G4JSF code base, given the evident cost of keeping it current and given the lack of other support from the JSF community. Maybe having a second baby also cut my energy level.

In order to avoid burnout I need to work on the projects that are most interesting to me, which right now is looking like prototyping some extensible programming language ideas that have been burning a hole in my brain for the last two months.

So, I’m going to set aside the G4JSF work and I’m going to leave the Seam/GWT/JSF integration project in its current state. I am more than willing to work with anyone (at Red Hat or elsewhere) who wants to pick it up again and fix it up to work with Seam 2 (and JSF 2 when it comes along), but I can’t drive it further on my own. Anyone who was using that integration library and who is interested in fixing it, feel free to email me (rjellinghaus at gmail dot com).

This blog will shift focus away from Seam/JSF and towards programming language research and the experiments I’m doing. Hopefully that’ll still be interesting to my loyal readers — I’ll do my damnedest to make it so! Because I’ll tell you one thing, it’s pretty fascinating to me 🙂

Unless you’ve been buried under a boulder, you’ve heard about Android. It’s nice to know what my secretive Google mobile JVM guru friend has been working on all this time 🙂

The Android forum is awash in C / C++ developers ranting about how lame Android is because it doesn’t support raw access to the hardware. It’s too bad, really. I feel for them, and I agree: Android apps are going to be slow and clunky compared to native apps.

But Google’s making a play to build a whole new kind of mobile ecosystem here. Google’s betting on increasing mobile CPU power (certainly an excellent bet), and Google knows that if Android develops a personal-computer-like reputation for flakiness, it will fail and fail bigtime. I would argue that a big reason no mobile platform has really taken off yet (besides crazy licensing costs) is that native apps are simply flaky, and the more of them you run, the flakier your hardware gets. Since Android is trying to greatly increase the hardware diversity of Android devices, allowing raw native access combined with open developer access would rapidly drive the stability of the system to zero.

Native developers hate to hear this, because they spend their lives tuning for performance and chasing crash bugs. I’m sure many of them do write good solid apps. But I’m also sure that making native code crash-proof, especially across a wide spectrum of hardware, is very, very hard to do. And if you want to throw your platform open to anyone who wants to ship an app for it, you’re guaranteed that many if not most of those people will lack the uber native skillz to achieve that kind of cross-platform stability.

So HOORAY GOOGLE! for making Android a fully managed platform! Note also that Microsoft sees the writing on the wall; they’re working on a fully managed operating system, for crying out loud.

End rant. The most interesting thing I’ve yet learned about Android is that it is fundamentally trying to end-run around Sun’s IP restrictions on Java ME. Stefano Mazzocchi breaks it down in exemplary fashion. Super cool! It’s going to be really strange to see Sun and Google at each other’s throats, but given how close the two companies are, and given how heavily Android leverages Java-the-language while totally subverting Java-the-runtime, AND given how anti-patent-war Sun is trying to be, I hope for some kind of peaceful reconciliation there. Microsoft, on the other hand, will go apeshit. Let the fireworks begin!