Phone Conference P1450.3 Working Group
Friday, Mar 14, 2003, 1:15 to 2:30pm PST
Tony Taylor (chair & scribe)
We are in the process of reconciling the issues from the recent "on-official review" in these bi-weekly meetings. Reviewers are invited to participate (and would, no doubt add value to the discussion) as we go over the issues. The schedule for reviewing is identified in the list above. A new Draft-D07 will be created as the issues are addressed. A ResolutionDoc-D07 captures the key issues and decisions.
1/17 - Dan Fan / NP Test (reviewer comments incorporated)
1/31 - Dave Gallagher / Mentor (reviewer comments incorporated)
2/14 - Gregg Wilder / TI
2/28 - Bruce Kaufman / Teseda
3/14 - Don Organ / Inovus
WaveformDescriptions - Don suggests that the Min/Max statements to describe the edge constraints on a waveform be replaced by assert timing expressions. Tony took this idea one step further and suggested that these asserts could just go in the place of the timing expression prior to the event specification (thereby removing the extra braced statements entirely. This was generally agreed to, however, Greg pointed out that this is a different semantic than in the a Waveforms block in that it is defining constraint rules to be evaluated (not actual waveforms). This needs to be made clear in the standard. Tony has the AI to make this change to the draft for review at the next meeting.
Pragma - It was not clear to Don from reading the draft that a Pragma can be used anywhere that an Ann is allowed. Tony took the AI to clarify this in the STIL.1 draft. Note: The Pragma block has now moved to STIL.1, however it is still fully applicable to STIL.3 as originally defined.
Spec->Selector - Don asks why the TRC block allows a Category reference, but not a Selector. The discussion brought out that the original motivation for this was to allow the Min/Max values for a spec variable to define constraint limits. The syntax has move somewhat away from this concept now so as not to require a spec variable for a basic rule set. The min/max are definable in the TRC syntax in most cases (e.g., MaxSignals). It now is reasonable to add the Selector reference, and the group agreed to do so. It was also decided that the spec min/max way of defining limits is a good way to do things. Tony took two AI's: 1) to add Selector to the TRC block; and 2) to define the use of spec min/max for the case where a spec variable is used with no Selector.
WaveformDescription & Resources - Don questioned the purpose of the resource specification statement with the WaveformDescriptions block (e.g., NumberData, NumberIO). Jason pointed out that this set of statements would not work well for their tester architecture and a Pragma containing a model of the tester might work better in their case. Jason took the AI to create a TRC specification for the Inovus tester and see how much can (or can't) be specified in the current TRC syntax.
Tester Load/Run time - Jason introduced a new issue (not on Don's list) about what happened to the concept of specifying run times in TRC. Note: This was a discussion that Jason introduced early on, and was dropped due to: 1) Jason's lack of attendance; 2) lack of support for this feature by other attendees; and 3) general concern that such a specification would be too controversial, dependant on application techniques, etc. The wg has not totally disbanded this concept, but more ATE support for this is needed before we add it to the language. Note: The original proposal was to define test cost metrics which would be even more controversial (i.e., who pays what for which tester?).
Integer expressions - Don points out that some tester resources are approaching the 2G limit of an integer (as defined in STIL.0 as a 32 bit 2's complement number). Wg decided to propose a MaxIntegerSize statement to the STIL.1 wg. This would be added to the file header. By default the limit would remain 32.
A second question was how expression would handle compares where round off is concerned (e.g., is '100ns/3' equal to '33.33ns' ?). The wg decided that this would require a function call (i.e., round (x)) as part of an expression. The standard (STIL.0 and STIL.1) has avoided defining expression functions - it has left this to be defined as a user function.
Unlimited resource using -1 - It was not clear to Don from reading the draft that the -1 for unlimited resource applies to all blocks. This is to be clarified in the TRC intro statement.
PatternCharacteristics->PatternVariables - This statement caused Don much confusion. wg became equally confused when trying to explain it. The explanation confuses to capabilities - variables and parameters. To clear this up, the reference to macros and procedures will be removed as they are covered in other statements. This statement will define only variable usage and will allow to individually identify - specs, integers, signal-variables, and integer-constants.
All other comments in Don's reveiw - are accepted and appropriate updates to the draft will be made.
Date: Friday, Mar 28, 2003
Time: 1:15 to 3:00 PDT
Phone Number: 1-800-486-2460
AI1 - Tony - Update D07
AI2 - Jason - Create a TRC file for an Inovus tester
AI3 - Tony - Add MaxIntegerSize to STIL.1