Skip to content

Meeting minutes Wednesday

Jump to bottom
Erik Siegel edited this page Feb 6, 2019 · 11 revisions

Attendees

  • Achim Berndzen
  • Gerrit Imsieke
  • Erik Siegel
  • Norm Walsh
  • Ari Nordström
  • David Maus
  • Geert Bormans
  • Martin Kraetke
  • Jirka Kosek

Archive related steps

  • To standard step library!
  • p:unarchive
    • https://github.com/xproc/3.0-steps/issues/4
    • Find out what there is? Zip does not store the MIME types.
    • Step should read multiple files with include/exclude filters on filenames
    • XML files that have the wrong extension will now be read as application/octet-stream. You can try to de-serialize it later
    • The archives to process should now get the zip through a port
    • Add an option to override the content-type on read (so you can for instance force a file to be treated as XML)
    • You can specify you need the manifest as JSON
  • p:archive
    • https://github.com/xproc/3.0-steps/issues/3
    • Documents can be specified in the manifest as:
      • Specified in the manifest inline
      • Appearing on the source port (by base-uri)
      • Loaded from disk (like <p:document>). This happens when there is no document on the source port with the given base-uri.
    • Documents on the source port not in the manifest are ignored
    • Error when manifest refers to document you can't find
    • Manifest is mandatory
    • Output of the step is now on the result port
    • There is a report port on which an amended manifest flows out
    • add a parameters option for specifying options for the compressing (like passwords, etc.). Implementation defined.
    • The manifest entries can hold serialization options. You can specify defaults on the manifest root element.
    • If you want an XML file in from disk as-is, set the content-type to octet-stream or text.
    • You can specify the manifest as JSON as well
  • zip needs to be supported, everything else is implementation defined
  • What about https://github.com/xproc/3.0-steps/issues/27 ? Closed for now.

Validation

  • Validation steps all remain optional
  • Possible extra steps:
    • validate-with-json-schema
    • Validate-with-rdf-schema
  • xvrl reporting. Report port on any validation step emits this
    • Schematron validation still should emit svrl or xvrl, select with a parameter
    • Reports can be in JSON also
  • xvrl is still under development. Separate github repo for this.

File steps

  • head and tail are gone (now do this with text steps)
  • Steps remain optional
  • file-tempfile will become create-tempfile
  • Error handling needs some work
    • deleting something that isn't there is not an error
    • create a directory that is already there is not an error
    • touch creates a file if it isn't there. Also updates timestamps on directories if possible.
  • Move directory-list and load-directory-list here

OS Steps

  • All info like cwd and environment goes into os-info
    • Define the return XML structure
  • We added p:exec to the OS steps becomes p:os-exec

Grouping the standard step library

A + behind a step means it has a match option and is therefore likely to need a JSON equivalent.

  • XML steps

    • add-attribute + (for json: add-property)
    • add-xml-base
    • delete + (json also)
    • escape-markup
    • filter (json also)
    • hash + (json also)
    • insert + (json also)
    • label-elements +
    • make-absolute-uris + (json also)
    • namespace-rename
    • namespace-delete (new)
    • pack
    • rename + (json also)
    • replace + (json also)
    • set-attributes + (json: set-properties)
    • split-sequence (json also)
    • string-replace + (json also)
    • unescape-markup
    • unwrap + (json also?)
    • wrap + (json also)
    • uuid + (json also)
    • wrap-sequence (json also)
    • xinclude
    • xslt
    • xquery

  • Generic

    • cast-content type
    • compare
    • count
    • identity
    • set-properties
    • sink
    • tee
    • archive
    • unarchive

  • I/O

    • http-request
    • load
    • store
    • www-form-urldecode
    • www-form-urlencode +

About processing JSON:

SO?

  • Leave the steps as they are for now
  • None of us is very knowledgeable about this (yet), we need more info and intel

p:run category

  • Optional

paged media step category

  • xsl-formatter
  • css-formatter
  • Both optional

Naming issues

  • text- (head, tail, join) ?

Unconference day

  • Introduction by Erik (presentation based on the one for Saturday)
  • Achim demos and explains its JSON starwars pipeline
  • Norm demos and explains its markdown pipeline (when it works...)
  • Gerrit will explain something about XVRL
  • Demo of XProc editor by Marco (unless this will go to the demo jam)
  • Questions and comments (ask about JSON support specifically)
Clone this wiki locally