Xmlsh - a command language (shell) based on the philospohy of Unix Shells designed for XML

David A. Lee - Epocrates - Paper

Why? - Unix was a radical paradigm shift and 40 years of 'progress' have eroded the core design fundamentals. Data types are not byte/line streams,tools have not evolved with the data (XML), and working with XML is WAY too complicated.

Open Source/closed development - test it out, he needs the feedback but no coding assistance.
www.xmlsh.org

Pure Java, Saxon 9, Log4J, Optional external OS commands (not reinventing the wheel)

Why not write new commands? - The shell is aging....focus is on streams and byte streams...he wants to use XML natively.

He does a bunch of live demo...can't show that, but the link above does take you to the code. If you were here you could have gotten it on a tiny 512 USB stick.

Some annoying things but the one that seems to upset him the most is that console IO is limited in Java...no pure java implementation of clear screen...

Really what this is a personal project that everyone can try out. He has it working..but he'd love to see how it works for other people. Check it out.

Questions/Comments:
Murry Altheim: Has he looked into Groovey Scripting?
Response: more interested in a scripting environment...implementing a Groovy interface seems trivial. Something missing right now is the lack of documentation for calling it from another environment...but that is something for later.

It isn't ready to scale...and David is aware of that. He's hoping to learn something from other speakers about non-serialized XML.

Alex Milowski: Have you looked at XProc?
Response: He is looking for command line interface and XProc seems to be something that requires building entire documents for the XProc. But XProc to xmlsh seems doable.

Kurt Cagle:
Have you thought of integrating with eXist?
Response: I hadn't even heard of eXist until yesterday. but I definitively am going to look into it.

Another Attendee: Unicode support?
Response: It is a goal...since it is java it is native, but the parser is utf-8 but testing hasn't occurred on the extended sets. No fundelmental reason why it can't be end-to-end and I want it to be.

Same Attendee: Memory management?
Response: Xquery serialized to a native XML dbase. eXist might assist this problem.