Flutterby™! : 2000-06-08 17:57:59+00

2000-06-08 17:57:59+00 by Dan Lyke 3 comments

Alright! who sold Big Blue a clue?? I love fads like Linux :)

Can you repost his comments here? I'd rather not sign up for a mailing list just to read his one comment.

Dave said: Attributes are possibly optional hints to the processor, or things that make assertions about the structure below (like this is html, or this is an outline, or whatever). Don't build too much functionality in attributes because they cannot be extended, where structures of XML elements can be extended. As always, XML is only as good as the minds of the people who design formats with it. My response interspersed into his comments: > Attributes are possibly optional hints to the > processor, or things that make assertions about > the structure below (like this is html, or this > is an outline, or whatever). I'm don't think I disagree with this statement. One place I mildly disagree with your useage is in the file served by http://static.userland.com/weblogMonitor/logs.xml, since the URLs aren't a part of the information displayed to the user and are, in fact, pretty much atomic, I think they should be attributes and not data. ie: <weblog url="..." adimage="...">Weblog Name</weblog> The particular situation that set me off, however, was an XML file which specified field name translations. The actual tag name was the name to convert to (so this XML file had a gazillion different tags), and it had an attribute of the tag name in the source file that I'm converting from. Of course the source file could have a tag exist in several places ("firstname" could be a tag for several different types of people), rather than indicating heierarchy somehow the other attribute was a count of "how many tags of this name we expect to see in the source file before the one we want". So, instead of something like: <nameofweblogauthor comesfrom="name" location="1"> (with the location tag assuming that maybe we had a "<creator><name>" tag appearing in the file before a "<weblog><name>") This was absolutely screaming for something like: > <translation> > <target>nameofweblogauthor</target> > <source>weblog<source>name</source></source> > </translation> Which simply reiterates: > As always, XML is only as good as the minds of the people > who design formats with it. Bingo. In the longer view, there are no stupid technologies, only stupid implementations. One of the things that Perl has done *absolutely* right is built a culture around the language that makes it very easy and clear to anyone poking about that there are right and wrong ways to implement. I haven't found such a culture for XML (nor Java, my other current love-hate relationship), I've seen some glimmerings, but haven't run across the motherlode.