Re: [ebxml-iic] More COmments on TestCase material...

[Michael Kass <michael.kass@nist.gov>]

Jacques and all,

    Attached is the latest iteration of the ebXML Abstract Test Suite,
with changes reflecting this week's comments.  Changes include:

1) Change of URN to  'testcase' , not 'semreq'

2) SetMessage now changed to 'PutMessage'

3) Kept 'Party' column, to reflect possibility that a party other than the 
test driver
will be observable in the future.

4) Change 'XPath Expression' column heading to 'Test Message Expression'

5) Changed comparison operator in Test Message Expression from "="  to "==" 
for clarity

6) Separated "Verification" steps from "Correlation", "Put" and "Get" steps

7) Based upon discussion with Jacques and Monica this past week,
I changed ErrorStatus names to either "FatalPrecondition" ( any test step 
other than a conformance test step),
'FatalTest' ( a step status result where conformance verification takes 
place ) or FatalOption ( a test step that tests
an "optional" feature).   It is assumed that the implementation of these 
three status results could be further broken down at run time into:

FatalPrecondition.system  ( send/receive or other system failure for the step )
FatalPrecondition.notApplicable (undetermined or normal failure of test step)
FatalOption.system (send/receive or other system failure for the optional 
feature test step)
FatalOption.notApplicable (undetermined or normal failure of the optional 
feature test step )
FatalTest.system  ( send/receive or other system failure for the 
conformance test step )
FatalTest ( conformance test failure for the conformance test step )


I also sent Jeff Turpin an example XML Schema diagram illustrating how we 
could manipulate test message MIME headers
using XML declarative syntax.

Another possible point for discussion at the Monday conference call is 
actual implementation of a Test Harness, and the mechanism(s) that we might 
use for
message template manipulation.  Modified XPath, XUpdate and a message 
template language ( such as XSLT ) are three possible
candidate technologies that could be used.  I have not seen any modified 
XPath implementations for updating XML document templates.
I have seen open source XUpdate implementations.  I have also seen XSLT 
implementation of a template language would be
easy to implement. I believe that an MS test driver does not necessarily 
have to only use only one of these technologies. It could be that a 
"template language"
is the best way to implement MS testing... but XUpdate may be a better way 
to do Registry/Repository testing



Your comments are welcome,

Mike

Attachment: ebxml_ms_20_abstract_tests.doc
Description: MS-Word document