Archive for March 2009
As usual, day one back from a conference has disappeared in a miasma of hurried catchups, email rationalizations and a numbing sensation of generalized low-grade horror.
Day two must mean the summary of the conference experience. I’m looking at my twenty-four pages of notes with mounting distress at the level of detail. Some ruthless editing is in order. I’ve previously written about the Monday line-up, and followed with a partially incendiary post that spiked my stats and has probably ensured that I will have a crowd around me at my next talk. Remember, everyone, that pitchforks are sharp and flaming torches are hazardous items indoors.
Tuesday 24 March
At some point in the past, Kagan’s company had constructed an Eclipse tool for our service testing framework, so I was recommended to go along to this talk. I got to see about fifteen minutes of it before being interrupted by a call that I couldn’t defer. Unfortunately, this meant missing the demo part. Overall the subject matter looked interesting. It addresses the issue of constructing Eclipse builds and products, which, for some unknown reason is more difficult than it should be – mostly it’s the failure modes that are impenetrable. However, because of my vendor/consumer hat, the profusion of build approaches fills me with more trepidation than delight. You know the feeling – you’re watching the Cambrian explosion here, much experimentation of body plans and capabilities, and you are all positive but really you are gritting your teeth waiting for the Pliocene to arrive so you can use the stuff in anger. Conclusion – interesting. Next step – contribution to Athena maybe?
Koen Aers presented this session, wherein BPMN was hung within an inch of its life, cut down from the scaffold, eviscerated on the windlass, had its intestines burnt in front of its eyes and finally had its corpse quartered and dispatched to points north, south, east and west. My little joke. It wasn’t that sort of execution at all. Basically he showed how jBPM 4 can execute your BPMN-described models. He gave a quick intro to the charmingly retro little grayscale icons used by BPMN. Koen also said that the BPM/Workflow is a jungle of standards. I think he understated the case a little. It’s probably more like the particle decay cloud that emerges from the collision point of two positron beams, except a lot slower. Hopefully convergence will occur at some point in our lifetimes. Conclusion – informative. Next step – /me will leave this kind of thing up to the experts.
Here we go. This is the collection of short talks that I was curating. The point of these tiny talks was to bring people up to speed with what’s going on in the SOA Tools Platform project. Jerry spoke about the STP Policy Editor New and Noteworthy – not too much changed here, but some interesting elements have been added to help you edit Distributed OSGi remote services configuration files. Next up was /me with Enterprise Integration Designer: New and Noteworthy – this is where I got to practice being the bearer of bad news. This component is currently sleeping deeply, with no real effort being made to support it except by yours truly. A plea for help ensued, complete with cute puppy picture. Want to contribute? Take a look at the Wiki and get in touch! In a carefully-planned sweetener to disperse the bitter aftertaste induced by bad news, Vincent then presented SCA Tools: new & noteworthy. Now this project is really lively and is constantly adding new features. They’ve made some changes around the capability to extend the metamodel with new SCA artifacts and included an XML editor for defining composites and component types. Above and beyond the immediate application to SCA runtimes like Tuscany and Frascati, I think that this tool has great potential to help people construct composite applications in this brave new OSGi world we are entering. Finally, Adrian brought us up to date on Integration of SOA Editors in Eclipse using the STP Intermediate Model. There’s another big potential here as the basis for a SOA-style repository model. Conclusion – you tell me. Next step – this will be the subject of another post!
In this talk, Dietmar introduced us to some of the BPEL-y stuff that he’s been working on, based off the Eclipse BPEL project. If you view the stats, this talk got totally panned, with a lot of -1s. Two things come to mind straightaway. First, I think that the material in the talk wasn’t really quite ready for presentation. If an attendee wanted to go off and do some work with this, they needed to download and build the BPEL project first. That’s ok in my book, but many attendees might expect to be served their meal on a plate, rather than being handed a bag of raw meat and other ingredients. Personal taste applies, but know your audience. Second, the stats seem off. Last time I looked, I didn’t see any +0 votes in the stats for this talk, and there was at least one – mine. So I’m not so sure accurate reporting has occurred here. Conclusion – somewhat confusing. Next step – tighten up and have downloads.
I think there will need to be an aside soon on the Eclipse BPEL project. Not now though. There’s more sessions. No wonder we were all hosed at the end of the week.
This was a quick intro from Ricco about a nascent initiative to produce an industry-aligned working group in Eclipse about all SOA things. More on that as it gains more support and strength later. And, for those of you who are quick to turn around and say that SOA is dead, I have two words for you: Ker. Ching. That will be all. Conclusion – watch this space.
Last talk of the day. Rich and Markus lead us through the Galileo build, and dammit, the room is out the door again. Grr. Since I spend time looking inside the guts of this particular monster, I bailed early, before the funk started to rise in the room. Conclusion – I know where the bodies are buried already. Next step – keep shovelling.
At that point I retired to the bar for beer and sushi with some other Progressians. Having a distributed organization, and in an environment where the travel budget fairies have gone to the bad, conferences can provide valuable meet-up opportunities. The Wednesday writeup shall ensue…
In watching the presentations this year, it has come to my attention that some of them are boring. Yes, I know this will be an unpopular opinion, but I was startled enough by it that I thought I had to share.
I have always asserted that as a presenter, your primary role is to entertain. That is, you want to ensure that the arses stay on the seats until you get your point across, and when you do get that point across you want people to be engaged, with their brains on.
Remember, presenter – I have Twitter now. I can amuse myself if you are not amusing me. But haven’t I come for content? Sure, but if I enjoyed the presentation, then I’ll be happy with a bunch of recommended links.
Now unfortunately, there are some topics that are inherently dull and uninteresting in their own right, although this does vary with personal taste. For example, even the merest mention of BPEL gets me right into that day-dream about being out on the golf course on a sunny day.
Or, come to think of it, the one about chainsaws, which I won’t elucidate here.
You are kind of doomed with those type of topics – you had better hope that the attendance contains a cadre of narrow-niche hardcore heads who love it from their own particular perspective. But it will – unless you get the conference wrong.
Ok, so you have a topic thats intrinsic novelty has not put you at a disadvantage. Hypothetical example, Editing Tools for SmoochML Documents. The first mistake to make would be to say “First I will introduce the SmoochML standard”. Dude, if I didn’t know what it was, I wouldn’t be in your talk. You are a developer, so I understand your need to do obsessive corner-case coverage, but, you know, you’re actually off-topic. And then you might complain that you are running out of time at the end. Sympathy, I know thee not.
This is why I like short talks (although ten minutes is too short) – there’s no time for a gentle introduction, it’s off the cliff, into the sea, down as deep as you can go, and then pop out of the water panting and hollering. Forget beach entry.
If you are doing a fifty-minute talk, then I recommend you get in there with a good narrative. Set up a dramatic situation – what was the problem? What would be the consequence of failure? You must create several thousand SmoochML documents in 24 hours, or the Joker is going to blow up a hospital. And, all you have is three dozen red pandas that the programmers inadvertently left behind over the weekend. Yikes!
Next step is to resolve the conflict, tell the story, how did you deliver using the Editing Tools for SmoochML project? Just tell it. Then once it’s all over, go back over a few things and drill into them a little more. People will be more engaged – they may even ask some questions rather than sitting there breathing shallowly with glossy little eyes.
Oh, and go check out the stats for the talks and see if you can spot the ones where people weren’t engaged.
It’s 7am Wednesday morning in Santa Clara, CA. Breakfast is on its way and I’m trying to sort my schedule for the day ahead at EclipseCon 2009.
The last couple of days have been busy. Not as busy as previous years – there’s a marked downturn in the attendance due to corporate travel bans and a seeming go-slow in the US Immigration Dept responsible for issuing travel visas.
On tutorial day, Monday, I attended a PDE builds/Athena builder tutorial. Peak learning came from Andrew Niefer (the last PDE build guy standing) whos skilled explanations painted a picture of PDE build not so much as the be-tentacled monstrosity many believe it is. Top tip – think of PDE as a framework for generating Ant scripts.
Nick has slides and comments on his blog.
Verdict: early days yet, but potential there. Looks like most of the presentation to the user is going to be a minimized properties file plus a conventional project structure. It’s ok. Work to be done with tests etc.
I’d planned to attend the p2 tutorial in the afternoon, but turned up three minutes late (lunchtime visit to Apple store, natch) and the place was rammed.
Rant ensues – we do need to get sorted on room allocations. The p2 thing is popular, since it was forced down our throats last year, and it was in a small room. Downstairs, there were huge rooms with a dozen people scattered about. Grr.
In the heel of the hunt, I went to the Domain-specific Language Development talk, which turned out to be a good learner (the pain). I actually twigged how GMF works at last. The concern is that I may have lossed SAN points in the process – I’ve often remarked how learning the many Eclipse frameworks closely mirrors character progression in the Call of Chthulu RPG. What’s more, I learned about model tranformation with Operational QVT, which is good – will use it. I zoned out during the Xpand piece (jetlag).
Top tip to presenters – Address your audience, not your slides. The slides can’t beat you with sticks for a bad delivery. Ask your audience how they are doing now and then: are they following? Have they got the ellipse in the node mapping done yet? Do they need a minute to regenerate the model because they are newbies at reading your mind? That would be a good start.
To wind it up, I got fingered by The Powers That Be to do powerpoint karaoke session with the inimitable Mik Kersten of Mylyn fame (going to have to get that integrated with FUSE tools). We presented a new framework for identity theft in the cloud. ROFLMAOs there were in abundance.
Back to the schedule selection. At this point in my career I had expected to have mastered basic bi-location, but haven’t, so I thinking of relying on oscillating between sessions. Tuesday update later on. Don’t forget there will be tweetin’ ahead – #eclipsecon.