It’s just data

RSS based APIs

Matt Croydon: Translating that into an XML-RPC call hurts my head. That's because the MetaWeblog API doesn't specify how to deal with required attributes.  Or with namespaces.  Or with nested XML elements...  Of couse, one could one by one address these items, and in the process reinvent XML.

For reference, here again is a RESTLog post for this item.  Now here it is in XML RPC, with the troublesome attribute included as a comment.  Finally, here a hypothetical SOAP version, with the blogger API defined control parameters passed as a SOAP header.

IMHO, if one can find a way to work with transfer level authentication and authorization, one should.  If application level control information is required, then an alternate header mechanism may be appropriate.  And this need not totally be an either/or situation.  One can use transfer level authentication and authorization with SOAP.  Ideally, the blogid would not be a parameter, but would be included in the URI.  Things like that.

And, again, the offer is open: if consensus were reached around a weblog post API which either wrapped or subsetted (or both) RSS 2.0, I would work with others to provide the validator.

Update: The XML RPC example has been updated based on the instructions in the MetaWeblog API specification.

Update: An RFC has been issued which provides different guidance as to how the text of the source element is to be passed.


Sam, what do you mean? I'm at work on RSSLibJ (http://rsslibj.sourceforge.net) which will have a version ready for release soon - so RSS direction is important for me. But RSS is output, for RSSLibJ, so I'm not sure how it dovetails with what you're talking about.

Posted by Joseph Ottinger at

Joseph - Cool!

Here's some thoughts:

For the short term, an API that sent a single <item> via HTTP might be useful building block for anybody that might want to build a RESTLog client in Java.

It might also be useful to have a set of unit tests that use HttpURLConnection to connect to http://feeds.archive.org/validator/.

Finally, if others pick up on this suggestion and build weblog and trackback APIs this way, they will already have the Java APIs available.

Posted by Sam Ruby at

Sam and Matt, the MetaWeblog API spec does indeed tell you what to do with attributes, in fact it uses an example exactly the case that Matt says it offers no guidance on. And even if it didn't there's no cause to rip up the pavement and build a new road. We'll just add detail to the spec. I started a new page on the XML-RPC site. So let's figure out what to do, if you want to do that. If you want to reinvent the wheel, you can do that too.

Posted by Dave Winer at

Can we step back a bit and work out a problem scope and domain? If there's not a solution needed (i.e., one exists that fulfills the need, or the problem isn't a problem) that's fine - but since I'm working on a syndication library at the moment, it's a good time to address the flaws of the protocol, as well as addressing data transfer.

Posted by Joseph Ottinger at

Dave--

Thanks for clarifying. I feel like a moron for not seeing it before. I've spent quite a bit of time skimming and referring to that page, and somehow I missed it. After skipping right past it, I headed over to the RSS specs for the source element. Then I tried to figure out how to cram that back into an XML-RPC call. I think that's what hurt my head.

Oops.

I never claimed to be an expert anyway. :)

Posted by Matt Croydon at

Dave - rip up the pavement? Did Radio "rip up the pavement" when it *ADDED* support for the MetaWeblog API in *ADDITION* to the (classic) blogger API?

It seems to me that the RESTLog API is simultaneously both simpler and more powerful than the MetaWeblog API.

Posted by Sam Ruby at

Hey Matt, no problem, happens all the time. To Sam, no to both your questions. Radio still supports the Blogger API and recommend to others that they do so as well. In fact the MetaWeblog API is not a complete API without the Blogger API. Anyway, when you use all caps it sounds like you're yelling. If so, why? I don't get it. Anyway, I put up an RFC that addresses this stuff on the XML-RPC website.

Posted by Dave Winer at

Dave - I suggested that weblogging tools consider supporting RESTLog in addition to other API's. You asked "why rip up the pavement", so I restated what I was suggesting with emphasis on two words: added, and additional.

I'll take a look at the RFC and let you know if I have any questions.

Posted by Sam Ruby at

Sam - Thanks for the offer to include this type of validation into the RSS Validator. When RESTLog gets adopted across multiple platforms this will help interop greatly.

Posted by joe at

Completely specified or drop-dead simple.

Sam Ruby: Matt Croydon: Translating that into an XML-RPC call hurts my head. That's because the MetaWeblog API doesn't specify how to deal with required attributes.  Or with namespaces.  Or with nested XML elements...  Of couse, one c...

Excerpt from Blogging Roller at

It's Not Just Me

David Johnson in response to Sam Ruby's post about weblog APIs: It hurts my head too. I'd rather see a SOAP interface where everything is specified by WSDL, or a RESTLog-like interface where everything is drop-dead simple. I'd love to see a...

Excerpt from Matt Croydon::postneo at

For the HTTP API you might want to add a User-Agent header similar to the AppKey parameter of the XML-RPC API.

Posted by Mike Dierken at

OK, I've added a user agent to the example, but it is not exactly the same as the AppKey parameter of the original BloggerAPI...

Posted by Sam Ruby at

RSS based APIs. Matt Croydon: Translating that into an XML-RPC call hurts my head. That's because the MetaWeblog API doesn't specify how to deal with required attributes.  Or with namespaces.  Or with nested XML elements...  Of couse,...

Excerpt from Lasipalatsi at

A SOAP/RSS is simpler then you may think.

In another thread to my O'Reilly post on Weblogs, Web services and the future, an anonymous poster raises the notion of using SOAP instead of XML-RPC. While other disagree I think this notion definite has merit as part of a solution when combined... [more]

Trackback from tima thinking outloud.

at

Add your comment