Smalltalk Tidbits, Industry Rants

The first thing that leapt out at me when reading the OPML spec is that it doesn't even use XML namespaces. It was written in September 2000, of course, before namespace use became widespread, but still... a revision is definitely overdue.

You might also want to point out that Dave often uses OPML version 1.1. Where's the spec for that?

Clearly a better approach is to convene some "industry leaders" and come up with a heavyweight spec that never gets used.

Actually, Chad, the better approach is to actually run the spec by (either formally or informally) a set of other developers with domain-specific knowledge and let them make suggestions about where the proposed spec is unclear, unimplementable, or underdeveloped, and to provide a few workable testcases that future implementors can work with.

Either that, or you can go with Immutable Specs On Big-Assed Clay Tablets, Emerging Perfectly-Formed From The Forehead of Zeus.

Ayup.

quote a set of other developers with domain-specific knowledge and let them make suggestions about where the proposed spec is unclear/quote I wonder who are the experts with the domain-specific knowledge of shared outlining?

My jaw dropped when you mentioned that he actually wants to specify screen pixels, so I went and had a look at his "spec". OPML is crap, end of story. Let it die, and the sooner the better.

"Like everything Dave Winer has ever been involved with, the specs are all in his head, and it's up to the rest of us to figure out wtf he actually meant." XML-RPC is extremely easy to implement and the spec is clear.

Comment by James Robertson

Rogers - it's easy until you want to deal with various interesting character sets. Then - not so much.

I wonder who are the experts with the domain-specific knowledge of shared outlining?

So I take it that your money's on Zeus?

XML is not rocket science, though you'd never know it from some of the stuff aggregators are expected to consume. I daresay there's at least a small number of folks out there capable of making constructive suggestions about implementing an XML-based spec.

The presentational elements are in OPML for a simple reason: the creator of OPML needed them. They're supported in all of UserLand's software, Stan Krute's Java Outline Editor, and I think a few other places. I don't see how optional elements in an XML format reflect negatively on the spec. At most, they reflect on the format itself, but even then, is it a flaw of OPML that it contains optional elements you didn't need for blogrolls? If there are specific implementation problems you encountered supporting OPML, sharing them would be helpful. I personally have some concerns about the attribute-based extension mechanism, but I've haven't coded software yet in which it got in my way. A blanket condemnation of a format -- especially by your commenter whose only exposure to it is a skim of the spec -- is just noise. As for XML-RPC, exotic character sets are always a pain in the ass. I don't see why that should be considered a particular flaw of XML-RPC.

I don't know why anyone takes anything that idiot Scoble says with any seriousness at all.

He's a marketdroid from whose mouth spews nothing but spam and stupid blatherings. Sure there are tons of people that hang on his every word "because he is influential." He acts like some petulant 4 year old girl learning to be a floozy that she saw in one of her mother's soap operas; either gushing about some M$ "innovation" or blathering about anything negative said about some other product by any other company but M$.

As for his self-proclaimed "geek" status, I don't think so. It's all in his imagination.

OPML is imperfect. All specs are imperfect. OPML is simple. Many specs aren't. OPML was there first. Some of us have no interest in wheel reinvention. Quite the opposite: we like to add energy to a first mover. I seem to be humming songs about architecture astronauts, high and flighty, whizzing above the rest of us ...