Review of Test document draft

Summary: while still quite rough, I think this document is good enough 
for a first Working Draft. I also have some minor comments etc.

Proposal: that we publish
http://www.w3.org/2001/sw/grddl-wg/doc50/grddl-tests.html
as a first WD, subject to the editor making changes in response to 
comments on the WG list, at his discretion.

Detailed review:

Minor suggestions for Intro:

old text
[[
The tests are meant to satisfy the requirement for test cases covering 
all GRDDL features and library transformations as outlined in the GRDDL 
Working Group charter.  They should be used for testing the conformance 
of GRDDL-aware agents. The normative tests cover the required behavior 
expected of a GRDDL-aware agent.  The informative tests demonstrate 
expected behavior with respect to the issues resolved by the Working 
Group as well as other tests of robustness for software agents which 
consume source documents within a Semantic Web and generate GRDDL 
results. This document itself has (as a GRDDL result) a manifest 
document describing the test cases in RDF.
]]
==>
Suggest text ** to highlight changes
[[
The tests *address* the requirement for test cases covering all GRDDL 
features and library transformations as outlined in the GRDDL Working 
Group charter.  They *can* be used for testing the conformance of 
GRDDL-aware agents. The normative tests cover ** behavior *described
in the normative sections of the GRDDL specification*.  The informative 
tests demonstrate *other permitted* behavior with respect to the issues 
resolved by the Working Group. ** This document itself has (as a GRDDL 
result) a manifest document describing the test cases in RDF.
]]
Rationale:
1: 'address' is adequately weak, but bolder than 'meant to satisfy'
2: avoid 'should', 'must' except with RFC 2119 force (either explicitly 
or not)
3: Sentence [The normative tests ...] contentious. Unnecessary 
discussion, so modify it to an uncontentious sentence.
4: s/expected/other permitted/ because informative tests are not expected.
5: Delete: [[as well as other tests of robustness for software agents]]
    I don't think we have any.
6: Delete unnecessary: [[which consume source documents within a 
Semantic Web and generate GRDDL results]]

Suggestions on deliverables.

Put the zip at the end of the list so that the "The" in "The test 
driver" is a backward rather than a forward reference.
In the zip list, change "An RDF/XML serialization ..." to "The manifest 
file"
Add, near the top of the list:
   + A directory of tests on the Web
Change "A driver" to "An example driver" (rationale use of the driver is 
not required to run the tests)

Suggestions on Test manifest Format
Add the following paragraph.

[[
The output documents are read as if they would retrieved from
the URL of the input document.
Some input documents correspond to multiple output documents,
see below.
]]

Suggestion for Using the Test Driver
Add at end
[[
The tests do not require the use of this test driver.
]]

Suggestion for EARL Reporting
Add at end
[[
EARL reports for test runs can be generated in other ways.
]]

Suggestion for Namespace Documents and Absolute Locations
- delete 'much' in first sentence. I think it's true, but feel that the 
reader should be free to make their own judgement.

Suggestion for Informative Tests
Old text:
[[
These tests cover features not mandated explicitely by the GRDDL 
specification, but demonstrate behavior expected of a GRDDL-aware agent 
in the context of Web Architecture best practices (@@Appropriate 
[WEBARCH] link?). They also cover behavior suggested by the Working 
Group as a result of resolving certain issues.
]]
suggested text
[[
These tests cover features not *covered* by the *normative text* of the 
GRDDL specification, but demonstrate *additional* behavior *that* a 
GRDDL-aware agent *may exhibit*. They *reflect* behavior suggested by 
the Working Group *while* resolving certain issues.
]]

Received on Wednesday, 14 March 2007 10:31:28 UTC