Project Update #4

[peterbussch]

In this week's meeting, we spent the majority of our time attempting to solve a merge conflict which arose while we were rearranging our project repository to put it in line with the Obdurodon guidelines. Although the process of figuring it out took a really long time, we eventually fixed the problem with the aid of Dr. Birnbaum. Another issue we dealt with was the creation of SSH keys for our computers. This helps with affirming that your computer has write-access to a specific repository. Besides the Github problems, we discussed the markup for our corpus, in particular the question of whether some pieces of information are better encoded as attributes or their own distinct elements. We worked on the first draft of our schema, and made some changes in order to make it serve our needs more effectively. Besides all these issues, I've been reading and interpreting the letters, while Hunter has begun work on our HTML.

[djbpitt]

Whether information should be encoded as elements or attributes is sometimes important and sometimes not. The main syntactic difference is that elements can contain markup and attributes can't, so if you need to represent a piece of information that might have internal markup, you have to use elements. Another important difference is philosophical: attributes are properties of elements, while element content is … er … the content of the parent element. Whether something should be understood as a property or as content is not always self-evident, and sometimes there is no definitive answer. We find it helpful to think of the question as involving both modeling concerns (how do we humans understand the nature of the information?) and engineering concerns (what will be most straightforward for the computer to process when we query or transform our documents?). That second part will make more sense once we begin our XSLT unit.

Assuming there are no syntactic restrictions involved in a particular decision, you can use XSLT to transform attributes to elements and vice versa, so if you later think you made a sub-optimal decision, you can change your mind without having to retag everything manually. This, too, will make more sense once we've begun working with XSLT.