Clemens Vasters

Football. Cloud. Tech. Politics. Pictures. Principal Architect for Messaging/Eventing at Microsoft.

BloggerAPI, MT API, MetaWeblog API, Comment API, Pingback API, Trackback  ...  are you nuts?

I must admit that until last week I didn't really pay much close attention to all the blogging related APIs and specs beyond "keeping myself informed". Today I copied my weekend's work over to this server and now I have all of them implemented as client and server versions. Sam's and Mark's validator is happy with my RSS 2.0 feed and the experimental Atom (Pie/Echo) feed.

I have to say ... the state of affairs in this space is absolutely scary. Most of the specs, especially for the APIs are lacking proper information detail, are often too informal with too much room for ambiguities and you need to be lucky to find a reasonably recent one. Sam laments that people don't read specs carefully and I agree, but I would argue that the specs need to be written carefully, too. It also seems that because the documentation on expected behavior is so thin, everybody implements their own flavor and extensions and not only do the APIs have huge overlap, but it seems like any random selection of offline blogging tools will use its own arbitrary selection of these APIs in any random order. Since my implementation didn't "grow" over time, but I implemented it all in one shot essentially only since last Thursday and had to look at this all at once and what I found was just saddening. All of this has to be consolidated and it will be.

I am all for the Atom project and creating a consolidated, SOAP-based API for all blogging functions that the aforementioned APIs offer. XML-RPC was a good thing to start with but its time is up.  I am also for replacing RSS x.x with a spec that's open and under the umbrella of a recognized standards body and not of a law school, that's XML as of ca. 2003 and not as of ca. 1998, and that's formally documented (with a proper schema). What's there right now smells all like "let's hack something up" and not very much like serious software engineering. Ok, it's proven that it all works, but how about dumping the prototypes now?

 

Updated: