Second, the endpoints used to access an API are, at most, 1/3 of what makes up an API – the most important part of an API is the data passed to and also from those endpoints to do the tasks that the endpoints enable.

So sure WordPress and Tumblr have a Twitter-compatible API but what happens when Twitter changes how one of the endpoints work, like they did when they recently changed how since_id works or even how re-tweets work. At best the other sites will be able to make similar changes but the worse case scenario is that they won’t or cannot make the changes and then things start to get ugly. Anyone remember RSS 0.93? or even the blog related API’s to make posts from clients – all suffered from fragmentation.

That is not to say I am not excited to see the API pattern used by Twitter is being more widely adopted, it’s a well thought out and functional API – I just think that crowning it the King of micro-blogging APIs is a bit premature until the *whole* API, data formats and all, is documented.

And just like that, whoosh, we’re serving Atom-over-XMPP.

Nice job Twitter crew, sure enough in the body of the xmpp message is embedded the twitter post but formatted as Atom with all of the metadata goodness API consumers are just going to eat up.

Well now Jaiku has joined the ranks of Jabber-enabled apps and it has raised the stakes considerably with some great features.

Jaiku has for a while now supported channels and the bot will let you post and reply to channels as well as to your presence stream.

Nice job RalphM and the Jaiku team!

More bugs were slain, some new parameters were added and a general code tidying-up ensued – enough that we decided to push the release button on 0.3.

Details on the project are on the pyTwerp code.google page, details about the 0.3 changes are in the CHANGES.txt file and the download bits are here.

Woo! Go Open Source!

Anywho, because the patches were mostly one-liners and fixed bugs, I applied them and rolled out version 0.2 which can be downloaded from either the pytwerp code.google site or from my code library page for Twerp.

Decklin and I are comparing notes and swapping patches on some other changes to Twerp, so more changes will be coming soon. I offered to add him to the project so look for that to happen.\

FEED MANTRA: OPML is the TREE – RSS is THE VINE – CONTENT is THE FRUIT. TAGS and META DATA are the LEAVES

Except for me I would s/OPML/XML/ and s/RSS/Atom/ :)

I immediately thought of a couple of way to do that with Twitter :)

I could create a bot that people would friend and then send location updates – it would then keep track of events, messages, whatever for that location and msg people anything that matched.

hmm, hey Alex – whatcha think?

For me it was just easier to do without that data but I can see how Twittervision needed the info because of some of the extra features Dave added.

Update: Oh gnarly issue indeed – a comment in the meish.org article outlines that it was happening to people who didn’t even create Twittervision accounts but rather by people in their *friends* list who had accounts. Because anyone you have friended can see all of your posts Twittervision was able to also see them.

New phrase of the day: Unintended Consequences

So yesterday Dave Winer hooked his version control system up to Twitter. I was just idly wondering how long it would take Bear to hook all of CIA up to Twitter…

umm, about 35 total minutes :)

But I don’t even begin to claim that it was that complicated as I stood on the shoulders of giants.

When I saw that CIA.vc offered feeds of the commit messages it was just a matter of wiring Universal Feed Parser to my existing jabber bot framework. I like updating Twitter using XMPP as it allows me to hang other items off of the bot easily and implement them as simple commands that are sent to the bot.

So it’s all written but right now I’m not turning it on for the simple reason that CIA.vc generates *huge* amounts of traffic hourly and I don’t think anyone really wants to receive twitter updates for all of them. Here are the stats I copy-n-pasted just now from the CIA.vc status page:

4316 messages so far today, 4145 messages yesterday
13507 messages so far this week, 27222 messages last week
36635 messages so far this month, 120666 messages last month
1632723 messages since the first one, 1.46 years ago, for an average of 28.27 seconds between messages

I’m thinking of making it a direct msg bot that allows you to send it either a direct twitter message or an IM of the project you want to track and then it will send updates for that project to you.

Stay tuned

Wrong

It seems the Facebook API doesn’t allow for posting of anything except photos and from the entries I found in the developer’s forum, it’s an often requested feature that they are aware of.

Other little nits: during different initial sessions they ask for new information about your profile to make it more secure and one of them about creating a security question. No worries, I thought, easy enough — I picked “what was your first concert?” and typed in my answer (and no, this isn’t my answer for real :) but it seems they don’t have many people on their testing team that like the group “Yes” as it never took the answer.

It also never gave me any feedback that it was not a proper response. It just kept asking me the same question over and over.

anywho, just a quick note as to why I am thoroughly enjoying the openness of Twitters very flexible API.