Dave has updated his OPML Validator Beta in response to comments; among the changes are ones that clarify the difference between error and warning conditions as I discussed here and in the comments to the announcement and to the update. Thanks, Dave - I think these changes make the validator much more useful for developers looking to generate or consume OPML.

In the announcement post comments Dave said something importan, something that answers the question I asked here:

“All validators express the opinion of the person or people writing the validator. It’s always important to recognize that. People need to find out whether their OPML will work with other software without coming to me…”

This idea was rattling around in my head as I tested the validator, but I couldn’t quite verbalize it until Dave did it for me.

I still want to have a single source for validation of a format - so that users/authors/developers don’t need to go to a dozen different sites to verify that the output they’re creating is correct (at least lexically speaking), but it’s of tremendous value at least to have a place where I can check any OPML I might produce against Dave’s standards. In fact, Dave’s is the default validator I’ll use unless a “breaking bug” is found - with OPML still in its infancy Dave is the primary source for OPML truth at the moment.

In fact, in a way a proliferation of validators could be of great value. If each developer or organization that is processing data of a certain format gave others a way to check data against their specific criteria it would be fairly easy to troubleshoot your output without emailing individual developers. Of course this would only work if developers are willing to change when shown that their processing decisions are wrong. On top of that, all those developers would need a basic tool set to rapidly develop validators of their own, or else noone would spend their valuable time/money making something available elsewhere.

I also read Kevin Burton’s comments on the validator, and agreed with his concern regarding title and text (having experienced similar issues with Bloglines subscription list exports). That was until I read Nick Bradbury’s post and the conversation that he and Dave had in the comments. That was a wonderful, public gem of open development process; a joy to read!