Minutes of ISAC meeting held via telecom on 16 September 2004. Present: Peter Ashenden, Jim Lewis, Deepak Pant, Chuck Swart, Ajay Varikat Absent: (none) Next Meeting: 07 October 2004. TOPIC: Review signal update cycle changes imposed by VHPI force, deposit and release commands. Discussion: Peter summarized these changes. The fundamental idea is that the kernel has two variables for a signal: one for the driving value and one for the current value. The force and deposit commands change either or both of these variables under various conditions. ACTION: ALL to continue to review this, since signal updating is so important. TOPIC: IR2028: Clarify simulation cycle. Discussion: The author of the IR suggests that the LRM isn't completely consistent on the need to iterate between updating explicit signals and implicit signals. His suggested improvements can be incorporated into the VHPI changes to signal updates. ACTION: Peter to update the IR to reflect this. TOPIC: IR2037: Typo wrt NOW in the index Discussion: It was agreed that the LRM should treat the two references to NOW consistently. The proposed analysis was accepted. ACTION: Chuck to forward to VASG. TOPIC: IR2039: Minor typos Discussion: In anticipation of VHDL200X work, the claims should be reviewed and sent to VASG. ACTION: Chuck to send latest version to Peter. Peter to analyze. TOPIC: IR2040: Problems with OTHERS in aggregates Discussion: Continuing discussion from last meeting, it was agreed that the expression "or subelement thereof" in 7.3.2.2 a) refers to the formal, not the actual. It was also pointed out that replacing the term "subelement" with "member" in appropriate places may help resolve this issue. (Note that this replacement probably does not completely deal with the issue of slices in the formal.) ACTION: Ajay to prepare analysis. TOPIC: IR2041: Association of members is too restricted. Discussion: The author complains of three problems with association of members: 1. OPEN is not allowed with individual association. 2. Subelements should not be required to be with each other. 3. There should not be a requirement of an association for each member. It was decided that there were some technical issues, none of them compelling. a. For input ports, if some elements are OPEN and others are not, then the default expression must be evaluated and some of its subelements used, others ignored. b. Requiring subelement associations to be contiguous makes analysis easier. It was also discovered that some implementations accept certain cases of OPEN or unassociated subelement association. All known implementations require contiguous subelement associations, but some upstream tools produce non-contiguous subelement associations (in violation of current LRM requirements). It was decided that technical issues are not critical, so the real issues are usability: error checking vs flexibility. Therefore, the ISAC recommends that this be passed on the the VASG as an enhancement request. ACTION: Chuck to write up for (final?) ISAC review. TOPIC: Web-based IR system. Discussion: Bugzilla was discussed. Its strengths are its interactive capabilities. Notes and attachments can be associated with an issue. Good searching can be done. Dependency graphs can be generated. Its weakness are that it has some irrelevant information (such as target platform) and, to be useful, we must map some BUGZILLA state into more useful IR states. It appears that this mapping can't be easily automated. Another weakness is that we can't install this on the vhdl.org server. Peter has volunteered his server, but will need some resources (financial) to get this started. ACTION: Peter to make first cut at IR--BUGZILLA mapping. Chuck to send most recent Requirements doc to Peter.