Qumana was recently released as version
2.0. It looks like they've added a bunch of spam and advertising
features, but haven't done anything to address problems with their
actual infrastructure.
Sorry, I'm a little frustrated. This isn't the first time I've tried to
figure out Qumana's problems.
Early this year, Terry and I tried to work with them to resolve some
issues so that Qumana could be used with Conversant, but the person who
was supposed to work with me consistently failed to respond with
anything other than "we're crazy busy right now, can we do this
later?"
Terry wants to try again, so I took another look at the problem. I'm
writing this up as a form of documentation so that Terry can present
them with all the facts we have available. Qumana contains some good
ideas, so hopefully this will work.
The problem appears to be fairly simple: Qumana is truncating the
"blogid" to 15 characters. The blogid is a string of text which
uniquely identifes the weblog on a server. Some systems use numbers:
1234567. Others use names like "curiousgeorge," which may also be part
of the weblog's URL.
Unfortunately for Qumana, Conversant isn't like other weblogging
systems. Here the weblog isn't the entire site, it's just one part of a
site... and there can be more than one weblog on a given site.
Conversant's blogid's look like "truerwords;truerwords;index" or
"cognovis;terryfrazier;weblog/index". There are three parts there,
separated by semicolons. The first is the zoneId for the Conversant
zone which contains the conversation (site) where the weblog lives. The
second is the id of the conversation within that zone. The third is the
path to the weblog page within the conversation.
Unfortunately, Qumana is storing the weblog id's with only 15
characters. So the examples above are truncated to "truerwords;true"
and "cognovis;terryf", respectively.
It's doing all of this automatically. The users never even see the
blogid. When setting up a new "publisher" in Qumana, the user is shown
a list of weblogs on the server which he is allowed to edit (it asks
the server, by sending blogger.listUsersBlogs call via xml-rpc). Here's
an example of Conversant sends back to that call (when talking to my
test server):
<?xml version="1.0"?>
<methodResponse>
<params>
<param>
<value><array>
<data>
<value>
<struct>
<member>
<name>blogid</name>
<value>testing2;fuzzynavel;index</value>
</member>
<member>
<name>blogName</name>
<value>fuzzynavel/index</value>
</member>
<member>
<name>url</name>
<value>http://cyclotron.macrobyte.net:8080/fuzzynavel/index</value>
</member>
</struct>
</value>
<value>
<struct>
<member>
<name>blogid</name>
<value>testing2;medlinetesting;news</value>
</member>
<member>
<name>blogName</name>
<value>news</value>
</member>
<member>
<name>url</name>
<value>http://cyclotron.macrobyte.net:8080/medlinetesting/news</value>
</member>
</struct>
</value>
<value>
<struct>
<member>
<name>blogid</name>
<value>testing2;testing;blog</value>
</member>
<member>
<name>blogName</name>
<value>blog</value>
</member>
<member>
<name>url</name>
<value>http://cyclotron.macrobyte.net:8080/blog</value>
</member>
</struct>
</value>
<value>
<struct>
<member>
<name>blogid</name>
<value>testing2;testing;index</value>
</member>
<member>
<name>blogName</name>
<value>Bloggin!</value>
</member>
<member>
<name>url</name>
<value>http://cyclotron.macrobyte.net:8080/index</value>
</member>
</struct>
</value>
<value>
<struct>
<member>
<name>blogid</name>
<value>testing2;testing;news</value>
</member>
<member>
<name>blogName</name>
<value>news</value>
</member>
<member>
<name>url</name>
<value>http://cyclotron.macrobyte.net:8080/news</value>
</member>
</struct>
</value>
</data>
</array></value>
</param>
</params>
</methodResponse>
That's an EXACT copy of what Conversant is sending back. It lists a
number of weblogs: note that none of them show a truncated blogId:
they're all longer than 15 characters.
However, Qumana's "publishers.dat" file, where it stores the
information about the weblogs to which it can publish, *does* show the
truncated id's.
When it tries to publish to one of these weblogs, or get the category
list, or do anything else that requires sending the blogid, Conversant
throws an error because there is no weblog with the id it provided.
Until they fix this problem with truncated blogid's, Qumana can't be
used with Conversant. Period.