Wednesday, March 19, 2008

EclipseCon: Tuesday

The morning started a little late, having forgotten to actually turn on my alarm clock after carefully setting the wakeup time. Who would have thought I could shower, shave, and dress in just 10 minutes! In hindsight, it worked out okay to have gotten an extra hour of sleep; all these blogs take a long time to finish before bed. After sprinting down to the conference hall, I made it in time to Dan Lyon's Fake Steve Jobs talk.

He explained why he decided to do this blog. Reason one was boredom. He was tired of writing about IBM and EMC. Boring, he exclaimed! The second and most important reason was fear. The print media business is being disrupted by the internet. So what to do about it? He wrote his "Attack of the Blogs" article as part of his denial process. Finally he gave in and looked into getting involved in the internet half of his company. Rejection for being "boring and dumb" prompted him to try his hat a blogging. He was interested in Jonathan Swartz' blog and figured, hey, what if I pretended to be an important company executive on a blogging binge?

Why Steve Jobs though? Primarily because he was struck by the strangeness of Apple's culture, or rather the Cult of Apple. Their main religious gathering being Macworld was where his parodies started. He actually admires Steve as a fascinating personality with depth. The fact that Steve's also a weird, uptight narcissist makes for excellent parody fodder. After writing for a couple of weeks, he was shocked to find how much interest there was in his blog. Word of mouth on the internet spreads very quickly. Of course he was still anonymous at this point. Now how to make money from it?

The mystery of who he was helped generate interest. As a result, he was finally offered a column to write for How ironic is that, given he was working for Forbes? There's a lesson here for big corporations around the world: remember to check under you nose. Finally he was outed, which cut back some of his audience, but also grew a new one.

So the question remains: Why does it work? Part of the reason is audience participation. Anyone can easily and quickly comment on the content to provide feedback in real time. The ideas someone suggests can easily affect the blog content for the next day. Anyone can even have a dialog via the blog c0mments. It let him combine his technology reporting skills with the ability to comment quickly, in a humorous way without the usual delays. He characterized his blogs as gross, silly, with occasional eruptions of seriousness. He can twist the story and get a message across in a way he can't via a serious story. I.e., he can tell the truths he can never tell otherwise. His various image-based parodies were just hilarious.

After that I attended the Artem's "GMF: Exemplary MDA" talk. The sound technicians showed up late to turn on the microphone so that was frustrating for poor Artem. He described the techniques used by GMF to build their generation frameworks. He pointed out that EMF's JET templates have grown quite complex which makes them difficult to read. They're also quite difficult to extend, unlike XPand which has more flexibility via its aspect oriented support. Support for merging generated code with an existing target is important for producing a flexible architecture that allow models to be changed and regenerated.

After that I had a chat with Yves of Soyatec about his technology for mapping Ecore models onto a forms-based UI model. It's cool and he'll likely write an EMFT project proposal for it soon. Then I had lunch at the Modeling table with a very friendly crowd. And after lunch I went to Kenn's talk Open Source Meets Open Specifications: Eclipse and the OMG.

I was overhead explaining that no, OMG doesn't stand for Oh My God that's complicated, and then my words were taken out of context and thrown back at me. Kenn outlined the kind of modeling technology available at Eclipse and related them to specifications from the OMG. Kenn's a good speaker. Here he is with Xavier and Yves of Soyatec.

After that I went to Eike Stepper's talk Connected Data Objects - The EMF Model Repository. He explained some of the basic aspects of EMF's resource framework, the underlying mechanism that CDO extends, and talked about some of the limitations of XML, e.g., most implementations will load the entire XML tree in its entirety and then keep in all in memory. There's also no direct transactional support nor fine grained locking. In contrast, a database as a backing store has interesting advantages. CDO provides a distributed shared model where the server logically maintains a huge object graph and each client can work with a direct view of a subset of that model. Then he showed in a demo how to tailor a generator model to generate CDO compatible implementation classes and how to configure the server communication and repository. His demo went on to show the technology in action. There were lots of questions which I think reflected a great deal of interest. Over all, it's very cool and he's a great speaker.

After that I had a nice chat with Sven and Wolfgang about oAW and about what Itemis is up to.

Time flies when you're having fun. Soon it was time for Chris and I to do our unrehearsed talk. It was a small audience, but I'm told it went well. Even Eugene and Wassim seemed to enjoy it, and they're tough customers to please! Eugene didn't even complain about all the colorful pictures in the slides. There were lots of questions from the audience and some good two way discussions.

What I thought was incredibly nice was that after the talk, Bruce from LeapFrog and Tamar from Cisco but formerly from LeapFrog came up and gave me a gift of chocolate along with a card to express their gratitude for all the help I provided them on the EMF newsgroup.

Of course the gift was very cool, but as I reread the comments in the card now, I'm so gratified to know that timely help does indeed have such a positive impact on EMF's users.

After that it was time for the reception. Free food and drinks! Donald was shocked to get an invite to a NetBeans party.

I kid you not. There were apparently some NetBeans fans with sad empty little lives who needed to live vicariously off the Eclipse crowd.

Not many people came to the "meet the board members" table, i.e., no one came, but it was a good place to hang out. There was a little PDE mini gathering of Brian and Chris.

It was also a staging area for modeling people anticipating the Mega Modeling Mania BoF. Here are Jean, Paul, Pierre, Bernd, and Ed (who if you look closely is where a no whining ribbon).

I had a little chat with Nicolaus who had attended Chris and my diversity talk. He's hoping to get involved in contributing code by helping provide patches to fix bugs and wanted some advice.

The Eclipse community has so many incredibly nice people. It's not only kind of heartwarming but also highly motivating for me to keep doing more of the same. I like playing the role of ambassador; though I'm not always so politically correct, as soon became apparent at the Mega Modeling Mania BoF. There were more than sixty people, so it took three photos to capture them all.

We had a lively discussion as I managed to stir simple differences of opinion into open hostility. Just kidding! It was indeed a lively discussion though. Such a diverse group. Some folks are definitely focused on the most abstract concepts that underly modeling while others care most about something that's grounded firmly in a pragmatic Java-based implementation. And even that characterization stirred up discussion about dynamic EMF models not really being Java-based because they aren't generated in Java; I tired to explain that of course it's still Java-based, using things like Java strings, integers, and so on, but sometimes it's best to leave well enough alone.

The issue of associations reared its head, a precursor of Thursday's OMG discussions I'm sure. I tried to express my perspective that typically associations are just pointers in an object oriented language and that if they are more than this, they becomes objects in their own right, with pointers to the objects being associated. But again, I find not everyone likes to think in the same concrete terms as I do. Oh well, I swore I wouldn't leave the room with any work items like yesterday's BoF and was successful in that. On the way to the next BoF, I got the person who was asking me to add new things to talk to the person asking me to remove existing things hoping they would cancel each other out...

The UML Tools BoF had quite nice turnout as well.

The folks form Anyware want to get involved to contribute some of the cool stuff from Topcased and Papyrus.

We need to talk some more tomorrow about how best to enable this new group of commiters without impacting the UML Tools project's release schedule.

I was pretty tired after this, but decided to head down to the bar with a few remaining modelers. When I got there, I spotted Andy and Nicolaus in the bar and Andy gave me a quick Glimmer demo. It's cool!

Mariot and Jonathan were nice enough to buy a round of drinks for a bunch of us.

We were all pretty tired, so soon we were off to bed, or in my case to write the second half of this very long blog. I sure hope someone is reading it! It's a lot of work but I'm sure will help me preserve my fond memories of the week's events.

1 comment:

RĂ©mi Dewitte said...

"I sure hope someone is reading it!"

-> I do with interest. Keep posting :).