05 Jul 2002
Fri, 05 Jul 2002
there is no spoon asks about Radio to Movable Type
there is no spoon says: "It hardly matters what this guy writes, it seems bound to be worthwhile. And if you don't care for the content, just dig the presentation. I've certainly never seen a Radio blog this nice. (Nothing against Radio, but I'm so totally ready to jump ship like David Watson recently did. I just would like to know how he did it and his comment system doesn't seem to be working so I can ask...)"
First of all, THANKS for pointing out that my Movable Type comments were broken. I took enough time tonight to debug the system, had no clue it wasn't working, and discovered that indeed a redirect in the cgi was killing the comment system. Solved that by eliminating the redirect and now everybody should be able to comment till their heart's content. Let me know if you have any more problems.
Secondly, my migration from Radio to Movable Type was a long and studied affair. I did several test runs on a variety of hardware/software that I have here before deciding on a system. In the end, of all the ones I looked at including Radio, Nucleus, CocoBlog, Roller, and Movable Type, Movable Type was the winner when considering features, usability, ease of install and config, quality of documentation, support, community, etc. Hopefully, there'll be a day when all the blog content management systems use a broadly interoperable post format that makes imports exports between disparate blog systems trivial but we're not there yet.
I cannot emphasize how much usability was a factor in my decision. A good example? When I tried to edit my Radio templates, I would quite often hork them up so severely that the site blew apart at the seams (would load in the browser but looked like a circus side show after a tornado went through) and I'd have to resort to a backup. I've already done a ton of customization with MT's template system and in every case, those changes went without a hitch, looked great, and were easy to intuit. Note: HTML editing is not my day job.
That being said, I had a lot of help with my migration. Most of that help came by way of Jonathon Delacour. Jonathon provided perl scripts, written by his friend Herman Coomans. I hacked the perl scripts a bit to work with differences in my environment. I've since written Jonathon to express my gratitude and ask for permission to document further what the migration involved. Jonathon's doc and scripts are excellent, work great, assuming you know enough technical stuff to hack the scripts to your paths, linux vs. windows, etc. Unfortunately, I haven't received any response from Jonathon. At any rate, Jonathon was very gracious in providing his scripts and instructions via email so if you want to pursue this as far as I did, send email to Jonathon and ask him to send you the zip. Good Luck!
Posted at: 23:32 | permalink
Jon Udell: what about web services usage analysis?
Jon Udell says: 'Now I'll relay a question asked of me at a talk I gave recently. Paraphrasing: "If you have a rich client like Flash 6 front-ending web services, then where's the clickstream, and what do we measure?"
To the extent we package back-end interaction as services, we'll get a better (higher-level) record of that interaction. Rather than infer from HTTP logs, we'll see things directly in database and service-broker logs. Since these sources are diverse, there will of course need to be a syslogd kind of mechanism to aggregate them.'
One approach would be to use a RESTian style to access the web services such that clients are using HTTP GETs. This approach has the advantage of maintaining complete compatibility with existing log parsing tools such as webalizer when used with a server that supports Common Log Format (CLF). If XML-RPC or SOAP libraries are required at the client, another approach would be to just make the web services spit out CLF and then use a tool such as webalizer to do the analysis. If employed comprehensively, this approach could probably provide as much or more data about the usage of a Flash client as any approach currently being used. I believe this approach is demonstrable using Glue from the Mind Electric along with JBoss 3, my preferred environment. Unfortunately, I don't have access to the boxes to demonstrate that right now.
You can get an idea of what I'm talking about by looking at my referer reports from webalizer. You see the entry for http://www.whatdoiknow.org/mt/mt-tb.cgi? That's an XML-RPC call served by CGI, typically via Apache. The article that I wrote last night talked about using the RESTian approach with XML-RPC and CGI. If enough people ping the TalkBack XML-RPC interface on my server, we'll see some interesting data appearing in the webalizer reports soon.
Also, another approach to the problem that Jon describes with aggregation is to use a proxy model to centralize requests. This precludes aggregating the logging since all of the requests would go through the same application server or cluster.
Posted at: 09:58 | permalink
Using Movable Type's TrackBack feature from Radio Userland
This is an example of how radio users can interact with the TrackBack feature of Movable Type weblogs. This is by no means exhaustive, particularly with regard to optional XML-RPC parameters, but it does demonstrate what is possible. The corresponding Radio article is here.
Update: if you are new to using usertalk scripting in radio, just paste the following code example into a new post, but make sure that the editor is in source mode first, not rich edit mode. Hit post/publish and the trackback ping should occur.
Caveat Emptor: You must change the text after the url= entry to refer to the URL on your site. Otherwise, the trackback ping will be recorded to the url listed in the code below, which is a temporary radio blog that I setup. Make the change before posting your entry
I used the following code:
<% scratchpad.s = tcp.httpClient (server:"www.davidwatson.org",
path:"/cgi-bin/mt-tb.cgi?
tb_id=386&url=http://radio.weblogs.com/0110393/2002/07/04.html#a2",
ctFollowRedirects:"5"); string.httpResultSplit (scratchpad.s) %>
This makes an HTTP GET request against the XML-RPC server running on the Movable Type blog. When successful, that blog will record the trackback to the URL (in this case the permalink of the post you're reading). You can see the effect on the Movable Type blog here.
I thought this was cool because it allows trackback to be used on a broader basis, demonstrates the possibilities of interop between disparate blog content management systems, and gets radio users involved in the referer linkback game. All we need now is for one of the enterprising radio wizards to make a simple macro out of this and we're all set. BTW, it looks awfully RESTian to me. ;->
Posted at: 00:06 | permalink
