Phone Conference P1450.1 Working Doc Subgroup Thurs Dec 20, 10:00 am PST --------------------------------------------- Attendees: ---------- Peter Wohl Tony Taylor Jim Teisher Greg Maston Doug Sprague Agenda ------ Review the p1450.1 spec, draft D13, Dec. 3, available on the dot1 website, with respect to the internal-review-resolutions doc also found on the web, dated Dec 13. Issues ------ resuming at: TT8: desire to support offsets from start of the patterns, so added Cycle statement to the X. Addition was accepted, but some concern about the keyword. Change Offset to be "Tag_offset", and Cycle to be "Pattern_offset" to make them more pertinent to the intent. Tony took [AI1] to do this. DM1: Greg identified that this change to namespaces, to have the scanchain names as part of the scancell namespace, is fine, BUT is backwards-incompatible with dot-0. Peter identified that we should highlight this change in our spec if we do this. Tony thought this was stated in section 6.3, but could not be found, so I gave Tony [AI2] to see that this happens. The group accepted to make both changes identified here (remove inheritscanstruct and expand the scancell namespace to include scanchains). cellref_expr: the definitions in 6.3 needs to be made consistent with definitions in dot-0. Might be more appropriate to call this a 'cellref_list' rather than an expression. The references to use of '+' (and no support of '-') need to be removed since these constructs aren't used here anyway. [AI3] to Tony. Greg is concerned that removing the hierarchy constructs here limits the functionality of what can be defined as chain segments. Tony proposed that maybe we should move the hierarchy constructs defined in .6 (reason for removing the inherit constructs defined here), into this document, in order to validate the change to the name space being made here. Tony took [AI4] to get access to this group to the .6 constructs. DM2: adding synchronization-latch information into the scancell lists. Peter has had several discussions about this with several different sets of people. After some MORE discussion on this topic, the Working Group pushed these additions into .6 because this group doesn't see that these constructs affect .1 issues but appear to be DRC-type (design, netlist) issues which are related to .6. DM3 - accepted as is. RK1 - PatternCharacteristics block - MUCH discussion here. Greg does not see a value of this block in dot-1, because if it's defined in the Pattern block then it's redundant to the information found right below it. Issues about maintenance of this information in this context are also a concern - if the user manipulates the patterns so they no longer contain one of the constructs counted in this block, who's responsible for correcting this block? Tony proposed two resolutions: (1) remove clause 17 from this doc. (2) remove PatternCharacteristics statement from Patterns block here. -- Agreed at this point to remove both this block and the reference, with the disposition to put this block under p1450.3 (under the Enviroment/TesterConstraints block), and to discuss with the p1450.6 people what they can do. RK2 - request to move the SignalGroups ref from Environment into the NameMaps. Started discussion on this request, which became more a finger-pointing exercise. Greg is concerned that without justification for this request, if issues are raised during balloting about the position of this statement, then in all likelyhood this change would be undone again. Greg took [AI5] to follow-up in particular why he thinks the Environment block DOES need name resolution for dot1. Next Steps ---------- Next meeting in two weeks, Jan 3, 2002, same time and phone#. Continue working to the same Draft 13 for next meeting. AIs --- [AI1] Tony change keywords for cycle-referencing in the X stmt. [AI2] Tony to define stmt for scanchains-in-scancell-namespace incompatibility w/dot0 [AI3] Tony make sure section 6.3 is consistent with dot0 definitions. [AI4] Tony provide access to this group to the .6 proposal. [AI5] Greg document why the Environment block should support name resolution.