Sean McGrath, CTO, Propylon

Sean McGrath's Weblog.

Saturday, December 13, 2003
    Jython as scripting language of the year for the Java platform
Vote here.


posted by Sean 3:35 PM
[Link]
. . .
    Final post on XML USA 2003
Highlight of the last day for me was the session I chaired in which Henry Thomson talked about pipeline processing (Yay!) and Sam Ruby walked through the Atom syndication project.
I did manage to exchange a few words with Rick Jelliffe at lunch before I had to dash off early. Ricks presentation on quality in XML processing and the distinction between "plan" and "probe" techniques is required reading for anyone approaching XML processing or schema design.
Unfortunately, I had to spend Thursday in D.C. so could not travel to N.Y. with Walter Perry to hear Rick talk to the XML SIG of New York. February of last year I presented on the XPipe proposal for pipelined XML processing to the same group. It was my last visit to NY actually. The city is well and truly back on its feet now. Chinatown was just crazy with people and traffic today and yesterday evening there was a large party crown heading into downtown Manhatten from New Jersey. The big apple is buzzing again.

posted by Sean 3:03 PM
[Link]
. . .
    Hairy tale of New York
    "They have cars big as bars they have rivers of gold but the wind blows right through ya 'tis no place for the old."

In the big apple humming the best XMas song of all time.
Much fun was had playing with the traffic all the way from Philly and then fighting through the Holland Tunnel to the seat of our libations and somnambulations in the east village.
Spent some quality time in the company of eight miles of books in Strands bookshop. The have everything basically and cheap too, especially with the way the Euro is against the Dollar at the moment. Ireland is such a *()%^ ripoff for books and CDs. Amazon is great but you cannot smell it, if you know what I mean. Some times you need to get up close, turn your head through ninety degrees and get up on some steps to hit the high shelves. Its part of the fun. (Tip to Amazon, allow users to browse books as if they were on bookshelves. It will help invoke the nostalgia of the physical bookshop. Probably not much you could do about the musty smell...unless...naw....).


posted by Sean 2:49 PM
[Link]
. . .
Friday, December 12, 2003
    Images, History, Language
In this city of Philly it seems appropriate to quote the playwrite Brian Friel.
    "It is not the literal past, the 'facts' of history, that shape us, but images of the past embodied in language."
A suitable Wittgenstein-ish sentiment for an XML conference where the struggle is (as ever) primarily with language and context in the ongoing battle for self-describing information.
I have long harbored the belief that in document-centric XML there is a time lag between content origination and its ex-post-facto 'obvious' structure. We look at the facts of our historical creation (the content) through the medium of language, dressed up in angle brackets.[1]
Paraphrasing another Irish literary giant (Joyce: "a nation is the same people living in the same place") - an XML schema is what you can create when you have documents that are the same structure used in the same context.

posted by Sean 5:22 AM
[Link]
. . .
    More XML folk
I met Dare Dare Obasanjo for the first time yesterday and also settled a question Lauren Wood had about the tallest XML geek (in the room. and probably at the conference as a whole). Norm Walsh comes first, followed by moi, followed by joint XML Cup 2003 winner, Michael Sperberg-McQueen.


posted by Sean 5:13 AM
[Link]
. . .
Thursday, December 11, 2003
    XML 2003 - another day
Out of town all day at meetings so did not get to the conference at all. Got a nice photo of me standing next to Einstein though in D.C.
Yesterday evening Tim and Lauren hosted a soiree when I met another couple of folks I have known for years electronically but never met in person before - Eve Maler and David Megginson or example.
I missed James Clarks talk on namespace routing language but managed to catch some of the town hall session on validation which was interesting. I made (surprise) a suggestion that a pipeline architecture is an excellent vehicle for achieving powerful validation capabilities without resorting to more and more complex expression languages. I made something of a mini-rant on the subjects of performance too.
Seconds per document is not the interesting metric. The interesting metric is documents processed per second. Beware von Neumann's curse.

posted by Sean 8:26 PM
[Link]
. . .
Wednesday, December 10, 2003
    XHTML - what it gets you and what it doesn't get you
XHTML - a recommendation and a warning (but mostly a warning) is an ITWorld article about XHTML.
I'm a big fan of XHTML but in my work I often come across situations where too much is made of the XML-ness of XHTML.
If you want to get semantic structure into your documents you have two choices for XML markup. Either create tags (and preferably a schema). Or - and this is, I think, becoming more and more compelling as XHTML tools improve - tunnell structure into your XHTML using well thought out stylenames for CSS. With a bit of thought and with the help of XHTML's span and div elements, you can go a long way to semantic markup. The XSLT mapping from XHTML+Styles to generic XML (and back) is left as an exercise for the reader:-)

posted by Sean 5:07 AM
[Link]
. . .
    Context is king
During his keynote yesterday Jon Udell quoted from one of my ITWorld articles : "...the four-ness of four is the least interesting thing about it." A couple of people at the conference have asked me what article that quote came from and what it is about. The article is called The impotence of numbers and its all about context being the real thing that turns data into information. XML Markup does not amount to a hill of beans without context. Markup context takes a number of forms including internal (within the confines of an XML instance) and external (how an aggregation of XML instances are inter-related.).
The former can be severly diluted by thinking of XML as a notation for associating atomic types with strings. The latter is not something that XML itself gets involved in (with the exception of the general entity structure which, whilst in XML 1.0, has something of a quantum verisimilitude is the industry.)

posted by Sean 4:44 AM
[Link]
. . .
Tuesday, December 09, 2003
    Day 1 (for me) XML USA
Jon Udell's keynote was very interesting. Other talks patchy. Looking forward to James Clark who tells me he is going to load a 35Mb OWL ontology into Emacs/NXML tomorrow.
Bumped into some more old faces. John Cowan, Walter Perry. Put some new faces to names: Sam Ruby, Wendell Piez.
Rick Jelliffe is talking at XML NY tomorrow. If possible, I'll be going up. Walter kindly offered a give me a ride if his car trouble is sorted.
Finally : Relax NG is to XML Schema as Igloo is to Gotham City.
Yes, it makes no sense.

posted by Sean 4:19 PM
[Link]
. . .
    Philadelphia here I come
Actually, Philadelphia here I am. Sitting in a good hotel that has Marriot has its meta-class. Its block to block Marriot instances around here, all nuanced in different ways. Glorious WiFi included in room charge. Saw some familiar faces yesterday evening when I arrived. Simon St. Laurent, Edd Dumbill, Eric van der Vlist, Bob du Charme. Also, met R. P. Bourret for the first time.
Its all deja vu stuff. We where here 4 years ago for XML '99. I remember giving a XML processing with Python tutorial.

posted by Sean 4:45 AM
[Link]
. . .
Sunday, December 07, 2003
    Hiding complexity versus pretending it isn't there
Mike Champion makes a number of important points in this article but the one that struck me most was:
    "Beware of the argument that complexity isn't a problem so long as it is hidden.".

Amen.


posted by Sean 2:18 AM
[Link]
. . .
    The bear necessities of networking
teddyborg.

posted by Sean 2:05 AM
[Link]
. . .

. . .
Weblog Commenting by HaloScan.com