cgmo-webcgm message

Subject: assessment of 1.0 dynamic tests

From: Lofton Henderson <lofton@rockynet.com>
To: cgmo-webcgm@lists.oasis-open.org
Date: Tue, 24 Jan 2006 11:10:17 -0700

Here is an assessment of the dynamic tests of the WebCGM 1.0 test suite, to determine what needs to be done so that they are valid 2.0 tests. There are actually a lot of changes needed, most of which come from object behaviors -- we replaced the old 3 with new 12, and we changed the default.

We should have some dialog about these tests (email anyone?), and should put it on a telecon agenda.

FYI, and for ease of commenting in email, here is the Summary from the start of the attached assessment document:

There is one global issue that needs to be considered for all tests: the version number in the legend block.

Should we change 1.0 to "1.0 & 2.0" for each test that is valid for 1.0 and 2.0? Or just use "2.0"?
2.0 tests that are no longer 1.0 valid must be labelled "2.0" only.
What about 1.0 tests that are no longer valid for 2.0 -- e.g., the 3 old object behaviors -- should they be kept (w/ 1.0 version number), with appropriate documentation changes, to indicate that the deprecation requirements of 7.2.2 are being tested?
About linking-anyURI-BE-07 -- should we add another URI test, about the clarified URI/char repertoire rules of 2.0 (sec.3.1.1.4)?

Object behavior changes:

several tests must be modified, because we changed the default object behavior (1.0 was: full+newHighlight. 2.0 is: zoom+newHighlight).
3 tests of old object behaviors: remove? or keep (with appropriate documentation changes) to test the mandatory mapping rules for 2.0 viewers at the end of 3.1.2.4.1? (The tests would have ProfileEd 1.0 in the CGM, but would indicate that it's a "2.0" test in the legend block).
the new object_behaviors test is formulated in the style of the DOM tests -- interactive table driven to test lots of behaviors. Should it be kept like that, or integrated into the format of the 1.0 test suite? (Same question would apply to other new 2.0 static and dynamic, non-DOM tests.)
Note that there were five anticipated tests that were never built. Some, I can't remember what was the intent (e.g., the other two para / subpara, content model). Testing 'layer' is clearer (but ... what do you test?). Testing (IE) browser bug also is clear (but...do we want to do it?).

Six tests will need minor (editorial) adjustment to their reference images (left frame) or source HTML text (right frame), because they contain actual text from 1.0 spec, which has changed in some cases. (This is cosmetic, but ought to be done -- the text ought to reflect the 2.0 text, although the test does not normatively depend on that.)

Note the third bullet of "Object behavior changes". Our new dynamic, non-DOM tests are not in the style of this test suite, but rather in the style of the new DOM tests. That's technically okay -- we can still link it from the TOC. But do we want to spend the time and effort for uniformity? The same would apply to new static tests, like NUBS and Protection Region.

Regards,
-Lofton.

Title: Assessment of WebCGM 1.0 test suite dynamic tests for 2.0 applicability

WebCGM 1.0 Conformance Test Suite

Dynamic Module -- Assessment of 1.0 Tests for 2.0

Summary

There is one global issue that needs to be considered for all tests: the version number in the legend block.

Should we change 1.0 to "1.0 & 2.0" for each test that is valid for 1.0 and 2.0? Or just use "2.0"?
2.0 tests that are no longer 1.0 valid must be labelled "2.0" only.
What about 1.0 tests that are no longer valid for 2.0 -- e.g., the 3 old object behaviors -- should they be kept (w/ 1.0 version number), with appropriate documentation changes, to indicate that the deprecation requirements of 7.2.2 are being tested?

About linking-anyURI-BE-07 -- should we add another URI test, about the clarified URI/char repertoire rules of 2.0 (sec.3.1.1.4)?

Object behavior changes:

several tests must be modified, because we changed the default object behavior (1.0 was: full+newHighlight. 2.0 is: zoom+newHighlight).
3 tests of old object behaviors: remove? or keep (with appropriate documentation changes) to test the mandatory mapping rules for 2.0 viewers at the end of 3.1.2.4.1? (The tests would have ProfileEd 1.0 in the CGM, but would indicate that it's a "2.0" test in the legend block).
the new object_behaviors test is formulated in the style of the DOM tests -- interactive table driven to test lots of behaviors. Should it be kept like that, or integrated into the format of the 1.0 test suite? (Same question would apply to other new 2.0 static and dynamic, non-DOM tests.)

Note that there were five anticipated tests that were never built. Some, I can't remember what was the intent (e.g., para / subpara, content model). Testing 'layer' is clearer (but ... what do you test?). Testing (IE) browser bug also is clear (do we want to do it?).

Six tests will need minor (editorial) adjustment to their reference images (left frame) or source HTML text (right frame), because they contain actual text from 1.0 spec, which has changed in some cases. (This is cosmetic, but ought to be done -- the text ought to reflect the 2.0 text, although the test does not normatively depend on that.)

Table of Details

linking-basicH2C-BE-01	Okay for 2.0.
linking-basicC2H-BE-02	Okay for 2.0. (But ensure that HTML text snippet in left window is accurate 2.0 text (sec.3.2.2.3))
linking-basicC2C-BE-03	Okay for 2.0.
linking-basicC2P-BE-04	Okay for 2.0.
linking-selectId-BE-05	Must fix default objBehavior, both in the reference picture and the bottom-frame description ("Operator Script"). In 1.0 it was essentially full+newHighlight. Now it is essentially zoom+newHighlight. ProfileId must be changed to 2.0. (Should OS mention different default behavior from 1.0? Yes.)
linking-selectName-BE-06	Must fix default objBehavior, both in the reference picture and the bottom-frame OS/description. In 1.0 it was essentially full+newHighlight. Now it is essentially zoom+newHighlight. ProfileId must be changed to 2.0. (Should OS mention different default behavior from 1.0? Yes.)
linking-anyURI-BE-07	Must update OS reference to WebCGM 1.0 and RFC-2936. Default objBehavior on 2nd object's link is now zoom+newHighlight (was: approximately full+newHighlight) -- fix OS description accordingly.
linking-multiLink-BE-08	Okay for 2.0.
behavior-picBlankC2C-BE-01	Okay for 2.0.
behavior-picReplaceC2C-BE-02	Okay for 2.0.
behavior-picBlankC2H-BE-03	Okay for 2.0. (But ensure that HTML text snippet in left window is accurate 2.0 text (sec.3.1.2.2))
behavior-picTargetC2H-BE-04	Okay for 2.0. (But ensure that PNG images of text snippet in left window is accurate 2.0 text (sec.3.1.2.2))
behavior-objHighlight-BE-05	Tests old objBehavior 'highlight'. Must be changed (at least the initial text in right window and the OS/description in bottom window). QUESTION. Should we keep this as a test of old-behav. mapping requirements -- and modify accordingly to clarify its status and purpose -- while adding the new-behav. tests (object_behaviors)?
behavior-objHighlightAll-BE-06	Tests old objBehavior highlight_all'. Must be changed (at least the initial text in right window and the OS/description in bottom window). QUESTION. Should we keep this as a test of old-behav. mapping requirements -- and modify accordingly to clarify its status and purpose -- while adding the new-behav. tests (object_behaviors)?
behavior-objViewContext-BE-07	Tests old objBehavior 'view_context'. Must be changed (at least the initial text in right window and the OS/description in bottom window). QUESTION. Should we keep this as a test of old-behav. mapping requirements -- and modify accordingly to clarify its status and purpose -- while adding the new-behav. tests (object_behaviors)?

fragment-idC2H-BE-01	Okay for 2.0. (But ensure that HTML text snippet in left window is accurate 2.0 text (sec.3.1.2.2))
fragment-multiPic-BE-02	Delete test.
fragment-fiveForms-BE-03	Modify test for 2.0: replace old objBehavior 'highlight' in 1st link with full+newHighlight; in 2nd link, there is default objBehavior, which is now zoom+newHighlight. Change OS/description of 2nd.
fragment-browserBug-BE-04*	(test never built)
interact-pick-BE-01	Okay for 2.0. (But fix so that HTML text snippet in left window is accurate 2.0 text (sec.3.2.1.1))
interact-pickRegion-BE-02	Okay for 2.0. (But fix so that HTML text snippet in left window is accurate 2.0 text (sec.3.2.1.1))
interact-screenTip-BE-03	Okay for 2.0.

otherAPS-para-BE-01	Must be changed because uses default objBehavior. (Either change to explicit full+newHighlight, or change Reference Picture and OS/description.)
otherAPS-para-BE-02	Must be changed because: 1.) uses default objBehavior (either change to explicit full+newHighlight, or change Reference Picture and OS/description); and, 2.) uses old objBehavior view_context.
otherAPS-subPara-BE-03*	(test never built)
otherAPS-subPara-BE-04*	(test never built)
otherAPS-layer-BE-05*	(test never built)
otherAPS-contentModel-BE-06*	(test never built)