ISAC Minutes 10 April 1995 Vantage Analysis Systems Chair: Clive Charlwood (crc@synopsys.com) Minutes: Daniel Barclay & John Willis Attendees: Clive Charlwood Daniel Barclay Jacques Rouilliard Bill Paulsen Andy Tsay Alex Zamfirescu John Willis Clive brought up his preference to present a chunk of IRs at DASC meetings at the DAC; about 25 IRs. Clive wants to get/keep momentum; how much can we each participate? How much do we want to? Work is involved writing up IRs and slides. (DAC- June 12) and (DASC- Friday after DAC?) Discussing dates: maybe before DAC, may be Thursday before DASC meetings (15th). Looking at IRs that need to be written; Action Item: Daniel will maintain a file that is a list of officially approved IRs. The IRs will have three states: not approved, ISAC, VASG approved. (1) create it (2) maintain it. Action item: Clive: to update action items. Items are to be listed with minutes of this meeting. Alex argued against this system (was concerned with multiple people updating the file). Vote: 5 for, 1 against: update the file method (versus email) Vote: 6 for, 0 against: mark lines as done (as opposed to deleting line) Action item: Daniel: Re-analyze IR1026 (basically expand one sentence into paragraph mentioning that it doesn't apply to VHDL-93). Action item: Daniel: Close out IR1026 or mark as not an issue. Action Item: Jacques: Will look at IR1010 and analyze Resolved without a vote at the meeting: Chair will keep a list of allocated IR numbers and will update IRlist file ASAP after meeting. Action Item: Clive: Will call Paul, Victor and Oz regarding their participation. Action Item: Alex will send Clive information on maintaining the ISAC mailing list. Vote: John moved to congratulate Alex on his maintenance of the mailing list. The motion passed anonymously. Paul and other noted that old IRs still are not readable by the world; Clive noted that the email archive does not work. Clive moved: Clive will be given ability to manage the ISAC mailing list and account. The motion was seconded by John. John clarified that the vote does not compel Alex to spend the estimated 2 hours required to explain the mailing list maintenance. The vote passed 5 for, 0 against and 2 abstentions. Clive moved: make all IR files world readable and include a README file saying "Draft, don't worry about spelling errors, etc."; John seconded (note: this is independent of WWW access to IRs). The motion passed 6 for, 1 against, 0 abstentions. Action item: Alex will make files accessible. Action item: John will clarify action item list from the last minutes. Vote: Each ISAC member will take about 2 days to write up IRs for presentation at the DASC/VASG meeting. 6 for, 0 against, 1 abstention. Proposed date for ISAC meeting is the day before the DASC meetings, the 15th or 16th of June. Action Item: Clive will set up a venue. Paul and Victor's work is to be divided: Action Item: IR1010 goes to Jacques Action Item: IR1017 goes to Alex Action Item: IR1026 goes to Daniel Action Item: IR1069 goes to Andy Action Item: IR1075 goes to Bill Action Item: IR1076 goes to Alex Action Item: IR1080 goes to Andy Action Item: IR1085 goes to Andy Action Item: John will cover Al Dewey's work items Daniel has noticed several formatting irregularities in the IRs: Some numbers are never actually used. Clive proposed that such numbers have trivial analysis and approval; All agreed. What is `93 disposition is N/A for new IR. Disposition rationale should be the same. What is date format; Month should have first letter capitalized; Daniel will move date to ISO date format such as 1995.04.10. Revision numbers get automated if we use a symbol. Action Item: Daniel: Update IR template & instructions following the decision above. Discussing IRs: IR1062: Type here must mean "base type" (because exact subtype match would be much too restrictive). Section 2.4, paragraph 2, S1 "type" means base type (because subtype would be too restrictive). Much discussion about little. Discussion of final resolution (i.e. disregard whether we re-interpret 1076-1993). Resolved: Type means "base type" in the two cases in Section 2.4; at least the future result type should be constrained and "convertible" to the signal subtype. Resolved for 1993: No over-riding of the LRM. Action Item: Clive: Will write up analysis and recommendation for IR1062. IR1073: No change; agreed the LRM says what it means. There is no need to change the LRM. Actin Item: ?? will write up IR1073 IR1078: Discussion; Issues user's perceptions, likely errors; original intent (pointing into line buffer and moving pointer); user's perception of modifying some line object. Action Item: John owns IR1078; is to fill in rationale and analysis. Alex walked out. Alex asked why Clive seemed to be trying to give/assign the IR to John. Clive said, "Honestly (~because) I don't trust you"). Discussion about Alex and Clive's response that triggered Alex's leaving. Brought up by Clive to check appropriateness. Clive felt that Alex was arguing only one case and wanted to fold in the other issues; group generally supported Clive. Alex seemed unwilling to share what he said he had already written up; then said he had to clean it up (to remove proprietary aspects). Alex was arguing one view strongly; but there are three or four approaches to analyze; also we wanted to limit scope of this IR and not add in other issues. IR formatting changes from another sheet: Use normal system (field values, approval) for "never-mind" cases "1076-1993 Disp." field for new IRs is "N/A" "Disp. Rat." field for new IRs is "N/A" "Disp. Rationale" field for new IRs is "N/A" "Superseded" field for new IRs is by default "N/A" May be used if superseded even before new language version Date format: changed to 1995/04/10 format Use RCS version number; remove manual version number. Officially approved : IR1073 Version 3, 1995/4/10 (after typos fixed) IR1062: Discussion on Clive's write-up of analysis and recommendations. Action Item: Clive will revise and mail out. IR1079: Yes, appears to be a mistake in editing Action Item: Bill will write analysis to accept proposed resolution number 1; some discussion to verify coverage of cases; no problem seen. (note: recommended IR text format- don't indent first lines of paragraphs). IR1088: More discussion (same as previous meeting) result- refer first part of 1088 to 1028. Action Item: Jacques will write LCS to cover issues; IR1088 will be closed (with recommendation of "write LCS". Action Item: Jacques will update IR1088. IR1093: Action Item: John will ask (e.g. on comp.lang.vhdl) whether anyone wants to use replacement characters.; Then write up IR1093. Resolved: Replacement characters should be removed by next revision; no changes for `93. Will recommend not to use in `93. IR1064: Agreed no change for `93 except to clarify that "call" means "contains call"; future no change except to clarify wording. Action Item: Alex: will clean up IR1064. Date for next working ISAC meeting: end of July, try Portland July 27, 28. Action Item: Clive to confirm next full ISAC meeting (Portland 7/27-7/28) IR1034: Looks same as IR1073. Action Item: Everyone: Everyone should review his migrated IRs to check for incomplete migation to 1993. Agreed, same result as IR1073, no problem; no change; Action Item: ??? to close out IR1073 IR1035: Leave it alone for now. Action Item: John: (IR1035) Will look for cleaner way to define non-local access restriction and definition of pure. What is intended direction; try to come up with new formal definition of pureness and restrictions? IR1036: Result is well-defined. Range of anonymous base type is implementation defined. Should use "others" to be portable. Action Item: Andy: IR1036 will write up analysis. IR1045: LRM doesn't say how to tokenize input; look at Ada. Direction: Examples should be legal. Examples are qualified expression, review wording of Section 13.2 "an explicit separator is required". Action Item: Daniel: IR1045 Will try to analyze. IR1048: Action Item: Clive will analyze IR1048: Code isn't legal anyway by first sentence Need to clarify definition of "in the default expression" Result of evaluating it should be an error instead of undefined. Closing: Clive: "I encourage everyone to get to ISAC work as soon as possible (when you get back).