From: Don Organ
Sent: Tuesday, April 22, 2003 8:00 PM
To: STIL. 4 (E-mail)
Subject: stds-1450.4: Teleconference minutes 4/22/2003
Teleconference minutes 4/22/2003
Jim O'Reilly
Ernie Wahl
Don Organ
Jim Mosley
 
Reviewed some email Don had sent to the call participants (only):
Regarding Inheritance - there are four or five ways to ways to consider inheritance:
  1. Propagation of data values - ala STIL.0's InheritWaveform and InheritWaveformTable. There was consensus that this form could be applicable to STIL.4
  2. Polymorphism - a more strict object-oriented view (along the lines of Liskov Substitution Principle, late-binding, etc.). There was consensus that this form is not applicable to STIL.4.
  3. Similarity between static types - such as STIL.0's Signals and SignalGroups (either or both can be used in a SigRefExpr). Inheritance is not necessarily required in cases such as these (although such relationships can be modeled that way - such as the UML diagram on page 119 of STIL.0).
  4. Defining defaults - such as default values for TestMethod arguments. Didn't explore this, but I think there was consensus that this could make sense.
  5. Defining an interface (along the lines of C++'s pure virtual functions). Might be useful for defining a TestMethod's interface.
Don and Ernie to explore this further offline.
 
Result value of a TestMethod. This is the age-old question as to what does a TestMethod return. We discovered that this wasn't a well-formed issue, since we all agree that TestMethods may have any number of Out arguments. We did discuss four possibilities:
1) the result of a TestMethod (whatever that means) is a Pass or Fail.
2) the result of a TestMethod (wtm) is Pass, Fail, or something that means "the test did not successfully complete" (often called "Error").
3) the result of a TestMethod (wtm) could be Pass1, Pass2 (...PassN), Fail (and maybe "Error"). This was intended to allow a TestMethod to indicate one of N passing conditions - such as in speed-binning.
4) the result of a TestMethod (wtm) could be anything (any type).
The quick unscientific poll had #1 beinging preferred by EW and acceptable to JM. #2 preferred by JO and JM and acceptable to DD, #3 being acceptable to JM and DO. and #4 being preferred by DO and DD.
 
Again, Don and Ernie to explore this further - to try to figure out what the real issue is.
 
-DVO-
 
 
---------------------------------------------------------------------------------
Don Organ                                       Inovys Corporation
don.organ@inovys.com                  (925) 924-9110 x122
---------------------------------------------------------------------------------