COMP60370: SOA & More Coursework

All written coursework is to be handed in to the postgraduate office by the following Monday, 12:00 am. In addition, please email XML documents, HTML documents, schemas, stylesheets, etc. to me at bparsia@cs.man.ac.uk with subject "COMP60370 Coursework". This will help me to check your answers and guard against plagiarism.

Please note that Uli and I have decided not to have a mini-project. The coursework (this one included) will be weighted accordingly.

[20 marks]
Find an example (in the wild) of a GET and of a POST (be careful with these!) request/response pair. I recommend using Firefox with Header Spy and Web Sniffer (Header Spy is good for scanning headers as you browse around; Web Sniffer has nicer layout, less distraction, and better control). For each example, please submit

The URI + user agent sending the request
The action (or "operation") this request initiates (at an application level; you may include some of the HTML to support your case)
Using Web Sniffer (or similar), determine the behavior of the server for that URI under HTTP 1.0 (with and without Host header). What are the major differences?

[30 marks]
Give two plausible cases where URI opacity may be thought to be bad practice? For example,

does it make sense to infer that a retrieved document is an HTML document if

it was retrieved using a URI ending with "MyWebPage.html"

it started with the octets corresponding to <html>

is is parsable as a valid HTML document

but was sent with a Content-Type line reading: image/png

what if some of these conditions (except for the shape of the URI) were missing?
if you don't infer HTMLness...should you still prompt the user to save this file as "MyWebPage.html"?
(Note that this is one example. You need a different one as well.)
Defend or attack these cases as exceptions to the good practice rule. Are these exceptional or common cases?

[20 marks]
Suppose you have two media types (i.e., for HTML and PDF) where each, roughly, defines a fragment to be a page range. (In HTML, the page range is with respect to the current print @media and is determined by where the start and end of a <a name=... end up.)

Are such fragments equivalent across type? Can you tell from the current description? (Be specific.)
Suppose you are setting up a service for digitized microfilm. You'd like there to be one document per microfilm (so you can have a single, nice URI for each one). Each document may correspond to tens of thousands of high resolution scanned pages. Analyze the difference between a fragment based URI design (i.e., http://ex.com/microfilm4342#page20) and a query based URI design (i.e.,(i.e., http://ex.com/microfilm4342?pagerange=20). Be sure to include effects on clients, servers, and intermediaries.

[30 marks]
Compare the basic architecture of two (simple) wiki engines from a Web, Web technologies, and Web browser perspective: The original WikiWikiWeb (or any similar system) and TiddlyWiki. I am not interested in the software architecture of the programs, e.g., whether it uses MVC, or is modular. I am intereted in URI design, use of scripting, use of HTTP, HTML, whether it's "restful", etc.
[20 marks]
Take the example HTML for blogs from slide 13, slide 14, slide 30, and write an XSLT style sheet that converts them (assuming the format shall be regular in the obvious way of each example) in to microformat versions of themselves using hAtom 0.1. (The microformated feeds should generate roughly the same Atom as on slide 15.) (Make sure you test with examples containing more than one entry!)