Use cases

A survey of the use cases on my desktop boils down in the following
way:

Most pipelines are simple, linear sequences. For example:

  1. Parse XML
  2. Perform XInclude
  3. Validate with RELAX NG
  4. Validate with XSD
  5. Transform

Some pipelines begin with non-XML sources, so stage 1 reads
a non-XML source and produces XML:

  1. Read non-XML
  2. Transform

Some pipelines produce ultimately non-XML results:

  1. Read pim.rdf
  2. Transform
  3. Produce sched.txt

Some pipelines have several non-XML steps:

  1. Read diagram.n3
  2. Transform
  3. Produce diagram.dot
  4. Produce diagram.png

Some pipelines read several inputs:

  1. Parse 1.xml
  2. Parse 2.xml
  3. Transform 1.xml
  4. Transform 2.xml (also using the output of stage 3)

Some pipelines have sub-pipelines:

  1. Read 1.n3; transform it; produce 1.svg
  2. Read 2.n3; transform it; produce 2.svg
  3. Read doc.xml
  4. Transform doc.xml (incorporating 1.svg and 2.svg)

Some pipelines produce several outputs:

  1. Parse XML
  2. Transform (producing 1.xml, 2.xml, ..., n.xml)

Some pipelines need to process all the outputs from a preceding stage:

  1. Parse XML
  2. Transform (producing 1.xml, 2.xml, ..., n.xml)
  3. Transform 1.xml
     Transform 2.xml
     ...
     Transform 3.xml

Some stages have parameters (simple value parameters in addition to
XML input file(s)). I don't think I have any use cases where one stage
has to generate simple value parameters for a subsequent stage).

Although I can imagine wanting pipelines that handle conditionals and
errors, all of the examples I have at hand don't have conditional
processing and either succeed or fail.

                                        Be seeing you,
                                          norm

-- 
Norman.Walsh@Sun.COM / XML Standards Architect / Sun Microsystems, Inc.
NOTICE: This email message is for the sole use of the intended
recipient(s) and may contain confidential and privileged information.
Any unauthorized review, use, disclosure or distribution is prohibited.
If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.

Received on Tuesday, 20 December 2005 15:58:51 UTC