This is a follow up to my How Not To Design An Extensible API post from last night. In response to my post Simon Fell stated that he doesn't see much difference between a system where XML is passed around as a mechanism for exchanging data and one where objects that implement a particular interface are passed around. The difference is that one is a tightly coupled system while the other is loosely coupled.

More comments below.

 


 

Categories:

Simon fell has been talking about creating a common API for the several .NET framework based news aggregators of which my RSS Bandit is one. Simon originally started with a model where a plugin implemented an interface which allowed it to take action on an XML fragment containing an RSS item. After some pushback from some folks he moved to designing API where an RSS item was represented as an object. Don Box stepped in and refined the API.

As a .NET RSS aggregator author I dislike both approaches. However I could imagine implementing the former while disliking how limited it was but consider that there is 0% chance I'd ever implement the latter. Sam Ruby hints at why the latter is flawed, below I throw my $0.02 into the stash.

 


 

Categories:

Torsten has outdone himself. He's completely rewritten the RSS Bandit GUI and based it on the Magic UI library. Check out the screenshots at one, two and three. My favorite features of the new GUI are tabbed browsing, nested categories and the fact that I can customize the look of feed items with a custom XSLT stylesheet.

More below on where to get RSS Bandit v1.1 beta, the ladder theory of relationships and thoughts on blogging.

Poll: Dilbert Character You Are Most Like?

 


 

Categories:

Epiphany time: I just realized one of the primary things wrong with the W3C working group charter process. Working groups are chartered for short periods, usually 6 months to 2 years after which they must justify their need to keep existing or be disolved. Since people on W3C working groups are professional standards people this motivates them to keep rapidly producing specs regardless of the adoption rate or level of industry support for the last spec they produced so the working group stays chartered and keeps churning out specs regardless of how benefical or dentrimental it is to the industry. I am curious as to what the ISO folks do differently in their process that allows standards groups to stay around and rev a spec every 5 years such as is the case with the C++ standards group.

Lots of rambling thoughts below. Thoughts on blogging, Star Wars gangsta rap, the ladder theory of male-female relationships, and being the most popular Obasanjo on the Web.

 


 

Categories:

It's been a pretty hectic week. I have to be test case complete for a milestone by the end of next week. Filing bugs, running bleeding edge builds and arguing with devs. I'm going to miss this in the coming months.

Wrestlemania XIX was cool, I got to see Hulk Hogan live and the Rock kicked Stone Cold's ass. It was a logistics nightmare though. They only setup one big screen for all of Safeco field then obscured it from most of the crowd with a big WWE banner. The only people who got their money's worth were those who brought their opera glasses.

Click below to hear the latest new features in RSS Bandit, details of my TechEd 2003 talk, some War on Terrorism Iraq commentary, and blogs that keep it real.

Poll: Favorite Fight of Wrestlemania XIX?

 


 

Categories:

After months of delays the article Andy and I wrote about XML and Relational Databases for Dr. Dobbs Journal will see print. Here's a sneak peak at next month's cover and TOC showing our article being the odd man out in an issue whose theme is algorithms. We were aiming for the XML themed issue but didn't get the article to them in due time. So if you are interested in algorithms and the relationship between XML and relational databases you should pick up a copy of next month's Dr. Dobbs Journal when it hits the news stands.

More below on Wrestlemania XIX and how recent developments in the RSS world affect RSS Bandit.

 


 

Categories:

I should be getting ready for work but I just noticed the following quote on Cafe con Leche as the Quote of the Day.

when you nest one XML document inside another, you should nest it directly, without using CDATA. XML is specifically designed to allow such nesting. CDATA is saying "the angle brackets in here may look like markup, but they aren't". So if they are markup, why put them in CDATA?

--Michael Kay on the xsl-list mailing list, Fri, 7 Mar 2003
Michael Kay's claim is true in the general case but to extrapolate and say that XML is specifically designed to allow such nesting is bogus because it explicitly prevents this.

More below on nesting XML documents, XQuery in SQL Server and an update on RSS Bandit.

Poll: Should I Start Writing For K5/Slashdot Again?

 


 

Categories:

March 25, 2003
@ 11:58 PM

I'm back from Spring Break. Panama City Beach was no Girls Gone Wild video (although according to the girl who sat next to me on the plane Daytona Beach was) it was still quite fun. I assume our trip mirrors the typical Spring Break story so I won't bore folks with the details. Anyway hearing tales of how other people drank all day, had sex on the beach and got kicked out of clubs probably isn't what K5 diary readers want to read anyway.

A bunch of bunch of rambling thoughts below on MSN advertising tactics, Miguel De Icaza's blog, Mix CDs and rediscovering my favorite blog from a few years ago.

 


 

Categories:

This is probably my last blog entry until I get back from Florida so I decided to talk about two do post thoughts on the two topics that have been occupying my mind all day before I forget about them and never get around to posting.

Poll: Favorite Mechanism for Processing XML?

 


 

Categories:

March 18, 2003
@ 11:58 PM

Categories: