XmlDatabasesAndXQueryForAgileInformationSystems

From SPA Wiki

"I" learnt a lot from this session so my thanks to all those who attended.

Here are the notes I made afterwards.

Contents 1 Observations from the XQuery session 2 Tool difficulties 3 Language difficulties 4 eXist database 5 Session presentation 6 Learning approach 7 Case study online

Observations from the XQuery session

Tool difficulties

The simple IDE was too poor for happy experiences. Justified though this may be in light of the stage of development of the project, it lacks essential features such as

• Line numbering
• Save as
• create XQuery files
• copy directories and retain the files MIME type
• Find
• Search and replace
• Syntax directed editing
• Error debugging
• Saving from the query engine saves as XML

Also the ability to run a script and see the result in a window.

Bracket matching is helpful, but the font used has { and ( very similar and hard to read.

These features may be available via webdav but I wasn’t able to show that in use in say, Oxygen.

Language difficulties

Needed a better crib sheet (as I had planned but never quite finished)

Need better explanation of the axis abbreviations - perhaps to start with the full axis names and then show the abbreviations (and a crib sheet for them)

Working with sequences is different to working with an element - need examples of processing both of these.

This case study depended rather too much on date and time processing - would have been better with a simple listing of appropriate functions. In fact I never solved the problem with time zones which the switch to summer time revealed on Sunday.

Case study also raised issue of the absence of type information in the stored documents and the need to coerce values - so need to go through coercion (implicit and explicit) and checking of types. Perhaps this is an inherent flaw with this case study?

eXist database

Typed data would be more valuable - if data is loaded with a schema, could the typing be retained?

Security problems with REST model - need to articulate the problem with the REST interface which Peter Marks identified.

Locking and integrity

Session presentation

An additional sheet to give the installation instructions was needed - it was on-line and on the slides, but clearer if a separate entity.

Same thing with the exercises. Although these were online and in the PowerPoint slides on paper and on screen, a paper version which can be seen in its entirety and written on would be better.

Forgot at the end to ask for their views on:

• The language (e.g. the absence of objects)
• The tool set
• The scope of application

Learning approach

It occurred to be later that I could have used the test-environment (which I introduced only at the end) as a exercise solution framework (which was indeed one reason for developing it). With this approach, each exercise is set up as a test with an expected result, the test code could either be edited into the test script or executed from an external file. Compile problems may be more difficult to figure out this way, but it would be a more familiar paradigm for experienced agile programmers, consistent with the agile method, and a better way to specify the exercise.

Case study online

I will bring the online version of this case study up todate ASAP - an earlier version is here:

http://www.cems.uwe.ac.uk/xmldb/rest//db/chriswallace/buses/index.xql