1. When I was a toddler, I was blond; I had long curly almost white hair. I must have turned to auburn around the age of 10.
2. I am often diagnosed with a light form of ADHD by people familiar with other ‘cases’. I have these hyperactive moments, even spurts of hyper-concentration, and I must say I enjoy them. I get loads of stuff done while I should be doing other stuff, i.e. I practise structured procrastination. In my line of work, I think the advantages and the drawbacks of this personality trait even out. On the other hand, you wouldn’t want me as your accountant.
3. For some reason house music has the same effect on me as coffee, and I need a dose of both to get my day started. I like my coffee the Italian way, small and strong, and my house music the soulful way: deep, vocal or slightly discofied. Invariably, the kick on the 1 and 3, the claps on the 2 and 4 and hihats in beween make me feel better and put a smile on my face. One day I’m going to the Southport Weekender festival for a total house immersion therapy. Personal favourites: Frankie Knuckles, Mateo and Matos, Masters at Work, Traxsource … To see that my daughter is picking up on some of that, and that she really likes music like Jill Scott, Jamiroquai, Stevie Wonder and Amerie is a source of pride!
4. While I have no problems being alone, I sometimes need to see and talk to lots of people in a short time. That’s one of the reasons why I organized Barcamp Brussels and the Bloggers meeting, and why I volunteer for events like the Bronks theatre festival and the Marollen diner. If you ever wanna see me hyperactive, catch me on one of those events. It’s like I’m on drugs, but in a good way (I think).
5. A dream holiday I would want to have is the following: go to Formentera for 3 weeks and build my own bass guitar.
   Why Formentera? Because it was the idyllic location for one of my favourite movies, Lucia y el sexo (Julio Medem). It’s like the silent niece of Ibiza.
   Why build a bass guitar? Because it’s just thé coolest instrument that exists and actually making one out of a large piece of wood must be so rewarding.

So who will I pass this ‘Five Things’ meme to? Why, the LeWeb3 buddies, of course:
Tom “Saturday Night Fever” Raftery (UPDATE: here),
Simon “Broken Laptop Dude” McDermott,
Clo “Empress of pr0n” Willaerts (UPDATE: here),
Frederik “Mystery ex-girlfriend anecdotes” Devries (UPDATE: here),
Bart “We’re quite flexible” De Waele (UPDATE: here) and
Maarten “Dave almost kissed me” Schenk (UPDATE: here).
Slap! There you go!

No related posts.

1993 @ Ukkel
I arrive at Sopres, one of the larger direct marketing / database management companies in Belgium. Fresh from university (and 1 year of military service), I expect to see RDBMS everywhere and dive into SQL. Imagine my surprise when I see that, yes, there are a lot of Sybase SQLServer databases around, but the bulk of the work is done with something they call ’square files’ (see below). They have built a whole set of tools to work with those and by using them myself, I learn to appreciate the advantanges of the system (speed, mainly) and grow a fairly accurate intuition for things like queries, indexes and outer joins.

Square files

What is a square file – or as we called it in French: “fichier carré“? It is a plain-text file with fixed record length. It looks square (actually: rather rectangular) when you open it in a text editor. E.g. a file XYZ2006.288, containing 20.000 customer records of each 288 bytes, would be exactly 5.760.000 bytes (5,76 MB). If the file was called XYZ2006.customers, each program that processed such a file would look for a ‘descriptor’ file in the same folder customers.d that would not only indicate the record length (288 bytes) but also define the fields within a record (e.g. fullname: 40 bytes, address: 60 bytes, postal code: 4 bytes, …). To make the square file easy editable, the last field could be “LF: 1 byte” that contains a line feed, so that each record is on a new line.
The advantage of this system:

• random access: record #13455 starts at byte 3874753. If you would use a variable-length record (like comma-separated-values/CSV for instance) you would have to count through 13454 linefeeds first.
• human-readable: you could just throw any file into vi or another text-editor and browse through the contents. You could also process contents with standard tools like grep (text-search) or tail (last # lines).
• separate data and metadata: it’s a bit harder to manage, but it’s easier to work with. A data file with 1000 records has a byte count that ends with ‘000′, or something is wrong.
• no (primary) index: the records in a square file have a certain order and if it’s the wrong one, sort them on whatever you want. Working with these files, you quickly understand that a sort is an expensive operation and where you can economize for speed.

The sp-tools

Through the years they has also developed a set of tools to work with square files. They all started with ’sp’, hence the “sp-tools”. I quickly realised they had an alternative for each SQL statement you could make:

• sp-sort: sort a file on certain fields (ascending/descending, number sort, unique records)
• sp-merge: merge two files sorted on the same fields and get a sorted result
• sp-query: select only records that qualify a certain query
• sp-expr: add fields to an existing file and fill them with a expression of existing fields, record number, conitional values …
• …

The most interesting tool was sp-match, that was used to do joins: merge the contents of two files by matching on certain fields. E.g. you have a table of purchases (customer_id, date, amount) and you want to add the postal-code from the customers file: you match on customer_id and get (customer_id,date_amount,postal-code) as output. Using this tool and its variants (sorted/non-sorted match table, inner/outer/left outer) really taught me how joins work, what has to be sorted (indexed), what has to be in memory or on disk, and that is still the basis of my intuitive SQL optimisation skills. We basically did by hand all the stuff that SQL does behind the scenes.

Disk space

We were working on Sun ‘dumb’ workstations and on our servers we had an enormous disk capacity at that time: several GIGABYTES! Hard disk sizes were still measured in megabytes back then. I remember a discussion on whether to buy one ‘huge’ 500MB disk or rather ten 50MB disks, because the last option would be more expensive, but faster (divided over multiple disk controllers).
A typical project required up to 100MB storage space (*gasp*) so all temporary files had to be cleaned up ASAP. That’s why there was a system of max-age suffixes: a file that was called abc.xyz.5t would be automatically deleted after 5 days. “.2t” after 2 days. If you left large files sitting around without .Nt suffix, you got spanked when they detected it in the weekly storage report.
The output of our work would be delivered on floppy disks, if it was smaller than 1.4MB or else on tape. I’m not talking DLT or DAT here, I mean tapes on spindles, the real thing. No email, no USB-sticks, no ZIP-drives, no CD-ROMs. Those were the days.

Related posts:

1. Binary confusion: kilobytes and kibibytes When I created my Bandwidth Calculator, easily the most...

Co-founder Danny Hillis escorts me down a hallway that dead-ends into an old-fashioned red phone booth. The phone rings. He places receiver to ear.
“The blue moon jumps over the purple sky,” he says, and hangs up.
Suddenly, the booth becomes a door, swinging out to reveal a vast, open room filled with engineers, gadgets and big ideas.
from Applied Minds Think Remarkably (Wired)

I remember Maarten, Henry, Frederik and me, in the early days of Keyware (in 1998, I think), preparing a demo for Walter Debrouwer’s Riverland company. The latter wanted to impress his prospective client BP, and so he wanted a biometric access control to his ‘labs’.

We hacked something together with a hastily purchased badge-reader-annex-intercom, linked to a PC’s soundcard, running the first beta demo of our speaker authentication software (based on a Lernout & Hauspie technology). I think we even added the Visionics (now Identix) face recognition software we licensed, linked to a QuickCam webcam. So the system would recognize your face, recognize your voice while you pronounce your passphrase and then let you in when it was sure enough it was actually you. Wonderful when it works. And when it doesn’t, you can always explain about false rejection, false acceptance, and equal error rate. Maarten and me even wrote a white paper on the subject, but I can’t find that document back, only references (PDF) to it.

Frederik is now at Vasco, Maarten is at Imec, Henry has set up Broncoway. But I have no idea what happened to Veronique, An, Anke, Rudy and the lovely Julia. Maybe it’s time for a reunion.

Technorati: biometrics

Related posts:

1. Automated initial image tagging: Ojos Inc What meta-data do we have for the average digital...
2. IBC Amsterdam: bigger, better, faster I spent Saturday with Clo at IBC 2006 (Amsterdam),...
3. Geek dinner in Gent : the pictures Exactly what a geek dinner should be like: someone...

When you get in, you are screened by the “Pain Commission” to see in what stage of the process you are: Pain, Rage, Resent or Over. I got the black ‘OVER’ badge, because actually, yes, I’m over it. You also get 20 Pain Euro that you can use later in the event. The most important part of the Congress are the one-on-one conversations you can have with ‘pain experts’. These are an mix of 25 well-known (Jos Geysels, Pascale Platel) and lesser known people that you can have a 30-minutes conversation with on some aspect of love, pain, regret, guilt, … you name it. These experts also belong to one of the 4 groups mentioned above, and you can only book a time slot with a person of your colour. If you want to, you can however use your pain euros to sell your badge and buy one of another colour at the “Pain Stockmarket” (in some cases, you get paid for this, in other cases you need to pay: supply and demand and all that jazz). There was also some karaoke theatre going on, but I never got to that.

So how did it go for me: I got in and booked a ‘black’ date with Mieke Debruyne, journalist (on “missed opportunities” – how prophetic this would be) and another ‘black’ one with Mieke Deley – dancer (on “guilt” – no picture, sorry). At 21h30 I got asked to fill in a vacant spot, but that turned out to be double booked. At 22h00 I was supposed to have my conversation with Mieke Deley, but since our table was double booked, we waited for 15 minutes and then just started talking while sitting on the ground in a corner. The conversation was great fun, we lost track of time and since no one told us our time slot was over, I only got to Mieke Debruyne 20 minutes late, at which point she was already talking to some one else. The organisation was rather clueless (yes, using Excel sheets can be something of a black art). When I asked to maybe book Ms Debruyne at a later time, the lady in question seemed to have disappeared. But since talking to Mieke-the-dancer was so much fun anyway, we told every one of the Love Pain hostesses who interrupted us – and there were quite some – that we just started our session, which gave us a load of free drinks, half a dozen of different locations and about 2 hours of off-and-on sparkling and very open conversation.

End conclusion: there’s still some work in the details, but the idea is brilliant. The general mood amongst visitors is something between openheartedness, amusement and genuine interest in what’s going on in the minds and hearts of the other person. That goes for most of the conversations I had that night, not only the 1 ‘pain expertise’ that I managed to arrange.

If there is a new edition in 2006, I’ll be there! Oh, and if Mieke Debruyne would be so kind to drop me a line so we can finally have that session, I’d be so grateful.

No related posts.