show all comments


Grady Booch in damage control mode

December 16, 2005 02:02:55 +0200 (EET)

Grady Booch has swung into full damage control mode, with the main force of the attack directed against our CEO, Juha-Pekka Tolvanen. Back in October, Booch said on an OOPSLA panel that he typically throws models away, but never throws source code away. Juha-Pekka quoted him on that, and his quote was picked up by Microsoft's Harry Pierson and Steve Cook.

Now Booch has written in his blog, and commented in Juha-Pekka's and Steve Cook's blogs, accusing Juha-Pekka of misquoting him:
It's sad how one can be misquoted and then for that misquote to be picked up by someone else with both then making a spin of the events to support their position. How silly is that.

Let me just state for the record that Juha-Pekka is one of the nicest guys around, and would never purposely misquote someone. We're all only human though, so let's see if anybody else was in the panel and recorded what Grady said. As it happens, there are at least three: here they are, with their versions of Booch's words:

Ivan Moore: "I often throw models away but tend not to throw away the source code"

Brian Foote: There was something decidedly devolutionary in Grady Booch's overdue exhortation to cast models aside

Erik Rose: "I typically throw my models away, but I always save my source code"

So it looks like Grady has his work cut out now, having to go round so many people to tell them they've misquoted him... ;-)

I have no problem with Grady explaining more clearly what he meant, and indeed later in the panel he said he only throws some of the models away. But I do have a problem with unfounded accusations of misquoting. Let's hope Grady is as quick to admit he was wrong about being misquoted, as he was to admit he didn't really throw all his models away. Even if we disagree on technical matters, we can still be polite and accurate - whatever our guru status.


Recognition for MetaPHOR members

June 10, 2005 16:30:02 +0300 (EEST)

Those with a nose for history will already know that the roots of MetaEdit+ lie in the late 1980s and early 1990s in the MetaPHOR research group at the University of Jyväskylä. Two central members of that team were Kari Smolander and Matti Rossi, now respectively a head lecturer and a professor at Finnish universities. Their paper with Pennsylvania's Prof. Sandeep Purao was chosen as the best of the 490 papers published at this year's European Conference on Information Systems. Congratulations, guys! More in Finnish in Digitoday.

Kari Smolander (right) receiving the award for best paper at ECIS 2005

Kari Smolander (right) receiving the award.


Scott Ambler: OMG not doing a good job

May 04, 2005 12:33:35 +0300 (EEST)

Culled mercilessly from Scott Ambler's post in the Yahoo AgileModeling group:

What does the OMG really bring to the table any more? Their marketing brand? Remember, the OMG isn't a standards body like IEEE or ISO, they're merely a self-appointed industry group. Their ability to get people together? Seems to me that we can do that via the Internet. Their ability to attract the right people? Can we honestly say that the right people are involved with the development of the UML right now?
I'm not convinced that the OMG is doing a good job with UML and I think that it's time for a new approach.
- Scott

I have to say that my recent interactions with the OMG haven't exactly filled me with confidence. UML 2.0 still isn't out, and I hear there are over 500 open issues with the current draft. Talking to people at the top of the OMG, IBM and Eclipse I got the distinct impression that those people no longer separate the three entities much. It looks like we're back to the bad old days when UML was de facto whatever Rational implemented in Rose.

Yesterday I spent the morning with an expert on XMI and XMI[DI] (which stores the positions of diagram elements). The take-up of XMI[DI] is even worse than for XMI: despite its being released in 2003, only one UML tool supports it. If I tell you it takes that tool over 10KB to save the diagram position information of a single UML class with no attributes or operations, you'll probably understand why. A 20-class diagram takes nearly 400KB, of which 50KB is for XMI and the rest for XMI[DI].

Standards are great, but I think they work best when they arise from best practice based on good theory, not when some committee tries to create something without ever having done it in practice.


Viewing server logs in Excel

March 15, 2005 15:28:52 +0200 (EET)

The other day I wanted to check that our web server had been up and network connectivity OK through the day. One quick way of doing this is to glance through the WWW logs for any obvious gaps in time. Since that's boring and error-prone with big logs, I imported the day's log into Excel. Actually, I copy-pasted the contents, since that way Excel handles separating the data into columns automatically, without having to go through the import wizard.

Unfortunately, Excel doesn't allow graphing time data sensibly. If say there are nine entries at 1am and one at 11pm, it will show ten peaks evenly spaced throughout the day. Since I knew I'd seen time data spread out correctly in a graph before, I woke up the Office Assistant 'Clippy', who shook off the dust and actually answered my question right first time: Excel graphs date data correctly, but not time data.

Since time is relative, that gives us a chance. I added a new column, with values calculated from the time value*365*24, formatted as dates. This mapped the day's time data over 24 years, one hour per year. Graphing that with bytes-sent gave an instant view of the data, and after a little formatting as below, I ended up with the following: no real gaps, other than the early hours of the morning when most of our customers are asleep. Like I should be now...

WWW activity

Format details: Choose a 2D column view. Set the minimum and maximum for the X axis scale to 1.1.1900 and 31.12.1923, the base unit to days, major unit to years, and minor unit to months. Each major unit thus represents one hour, and minor unit 5 minutes. Change the number to a custom format showing just yy, i.e. the two-digit hour. For the data series, set the border to none and the area to black, and on the Options page set the gap width to 0.

Previous (14 total)