Bug 033: Alternative Serialisation

This is feedback on a Last Call Working Draft:

Developer Guide for Evaluation and Report Language (EARL) 1.0
W3C Working Draft 10 May 2011
http://www.w3.org/TR/2011/WD-EARL10-Guide-20110510/

Specifically § 5. Serializations of EARL Reports:

http://www.w3.org/TR/2011/WD-EARL10-Guide-20110510/#serialization

Currently this section is stubbed for further work. Why was no effort
made to bake in at least one subset or alternative serialisation as
part of the Last Call specification? It is a bug that no such effort
was made.

Looking at existing test framework systems:

http://en.wikipedia.org/wiki/List_of_unit_testing_frameworks

It is reasonable to assume that a generic test language ought to range
in expressive capacity from the very simple to the very verbose, if
not complex. This is covered to some degree in Bug 004. But subset and
alternative serialisations, on the syntactic front, accompany the
profiles described in Bug 004, on the semantic front.

Subset and alternative serialisations may have other ramifications
too, such as allowing the use of EARL embedded in other applications.
Consider, for example, EARL embedded in TAP or HTTP headers, for which
you'd probably want a flexible JSON format, or in xUnit XML, for which
you'd probably want a subset XML format.

The current requirements document is quote open on this front:

Requirements for the Evaluation and Report Language (EARL) 1.0
W3C Working Draft 29 October 2009
http://www.w3.org/TR/2009/WD-EARL10-Requirements-20091029/

For example it says:

“S01 / EARL 1.0 will be a vocabulary, the terms of which are given in
a set of specifications and explanatory technical notes for describing
test results.”

But it doesn't say that there will be necessarily any structure to the
vocabulary, a data model to go with the terms.

It also says:

“D01 / EARL 1.0 will define its vocabulary as a set of Resource
Description Framework Schemas [RDF]”

But it doesn't say that the serialisations, even their data models,
have to have anything to do with the RDF data model. This of course is
important if you're talking about a JSON serialisation, even if it is
possible to convert between it and RDF with 100% compatibility in both
directions.

-- 
Sean B. Palmer, http://inamidst.com/sbp/

Received on Wednesday, 11 May 2011 18:59:58 UTC