Saturday, February 08, 2003

The language gap

The language gap over on looselycoupled.com: "XML isn't a language, it's just a set of language rules."

Perhaps the biggest curved ball thrown in this industry ever is that XML is a "language". Boy if I had a dime for every time I've explained why it isn't a "language" in the way people normally interpret the phrase...

Friday, February 07, 2003

Tanga - a Pythonesque data notation

Tanga reminds me of two things. Some years ago now, I suggested something similar that I called Monty.
Same idea, take Python's brilliant insight that indentation structure (visual) can and should be used to unambiguously infer program structure (logical) - and apply it to structured data.

For a moment I harboured a hope that this language was a pun on the Irish word "Teanga", pronounced almost the same way, which means "language".

It turns out its named after a brand of underwear.

Zen, flow and emergence in information models

Jon Udell's musings about data emergence resonate with my own philosophy:

1. There is no such thing as the 'correct' information model.

2. Everything flows, data changes shape, it evolves, it emerges. You cannot - and should not - try to pin it down. It will refuse to be pinned down anyway.
In such a world, transformation is the only truth.

3. We should bulld XML systems so that all views of the data - ALL VIEWS - are potentially transformations of the underlying form. In my mind I see
an XML aware filesystem/DB. A filesystem/DB in which all accesses to XML files are - transparently to the user - processed through a transformational
pipe.

In the simple case, the transformation is the null transfomation. Then, as your need to mutate the data grows, you build in transformations at
the filesystem/DB level. I believe such a strategy would significantly extend the lifetime of any traditional, static, enterprise information
model.

It's odd but I think the biggest thing we have lost in the headlong rush to XML is the powerful RDB dualism between physical and logical
data structures. Come back database view technology. All is forgiven.

Tuesday, February 04, 2003

Pssst! There is no such thing as the perfect enterprise information model!

In the cut and thrust world of enterprise information management there is no room for perfection in information management. Even if it was attainable (which it isn't), there is no time in the project plan allocated to achieving it:-) The realities of enterprise information management.

Where did the term 'permathread' come from?

Leigh Dodds asks where the term 'permathread' came from.. Its a great word and occurs a lot on the xml-dev mailing list.

Monday, February 03, 2003

Rick Jelliffe (in fine voice) tackles infoset integrity, reliability and robustness

Go Rick!.

Its snowing

Snow dumping down (by Irish standards) at the moment. Time to curl up to a hot laptop and work from home I think.