RE: More issues with the tests

Stefan,
  I'll look at these. The tests that require the xpath data model are tests _of_ the xpath datamodel.  They're marked as optional and only platforms that support the xpath datamodel should run them.  (Similarly there are tests that require the ecma datamodel, and only platforms that support it should run them.)  There are also optional tests for the basic http event i/o processor that I will put up next week.  Only platforms that support it should run them.  

As for event.raw in test 178,  I had forgotten that this test used it.  I thought it was used only for the BasicHTTPEvent I/O processor.  I was planning to put an explanation in the test plan that platforms that want to test the BasicHTTPEvent I/O Processor need to add an attribute to event that contains the raw message.  There's really  no way to test http message delivery without building some kind of extension (other VoiceBrowser specs have required special server-side components), and I figured that adding this extension was no more burdensome and probably useful for debugging anyway.  

But in principle, platforms that don't want to test the basic http processor shouldn't need to build any extensions,  so I'm open to rewriting test178 if anyone can think of another way to do it.

- Jim

-----Original Message-----
From: Stefan Radomski [mailto:radomski@tk.informatik.tu-darmstadt.de] 
Sent: Friday, April 05, 2013 8:56 AM
To: www-voice@w3.org (www-voice@w3.org)
Subject: More issues with the tests

Hi again,

I had my interpreter run every test in the IRP and there are some remaining issues [1]:

- test153: XSLT transformation creates a cond="Var1/text() <Var2/text() " attribute
- test178: There is no _event.raw attribute specified in the draft
- test226: Wrong filename extension .txml vs .scxml in dependency
- test302-304: datamodel attribute is missing
- test321: Contains a variable name "$_sessionid"
- test323: Contains a variable name "$_name"
- test329: Would you consider it in spec to assign to attributes within the _event object? It's a convenient place to store things we computed per event processing.
- test330: There are some namespace issues with only the finalize elements having the correct namespace
- test333-339: You test for null, I set these to the empty string - spec says to leave "blank"
- test436 + 441b: requires NULL datamodel
- test456: Expects undefined + 1 to be 1?
- test463-486b: require xpath datamodel
- test488: Expects _event.data to be the empty string, whereas I have it as undefined
- test537-547 + 555: require path datamodel

With the exception of 301 and 569, all others are passed by uSCXML for the ecmascript datamodel. There is also some mentioning of an exmode attribute with the SCXML element that is unspecified afaict.

Best regards
  Stefan

[1] https://github.com/tklab-tud/uscxml#test-reports

Received on Friday, 5 April 2013 13:10:45 UTC