ISAC Meeting Minutes, 25-26 March 1994 ====================================== Location: Cadence, Chelmsford, MA Attendees --------- Daniel Barclay, Compass Victor Berman, Cadence Clive Charlwood, Synopsys Alan Dewey, IBM Paul Menchini, Menchini & Associates Bill Paulsen, Viewlogic Jacques Rouillard, Ecole Superieure d'Ingenieurs de Marseille (ESIM) Chuck Swart, Mentor Graphics Andy Tsay, Fintronic Alex Zamfirescu, Intergraph Agenda ------ * Voting Mechanism Discussion. * IR Database Discussion. * Non-Text Databases. * Disposition of IRs of unknown status. * Several new issues relating to 1993. * Disposition of 1993 Comments forwarded to ISAC. * Other issues. Last Minutes ------------ By acclamation, the group thanks John Willis for his work in taking last meeting's minutes. LRM Status (Paul) ----------------- The final edits on the LRM have been made. The information is to be delivered to Paul today, who will examine the changes shortly and report back to the IEEE. Should no changes be found that modify the meaning of the language, the LRM is then ready for publication. Repository Status (Alex) ------------------------ Accounts set up for all ISAC members. IRs currently exist in 2 large files. The RCS directory is also set up. Use RCS. We now must decide what to do about the new IRs and how to distribute new IRs. Clive will set up a number-allocating mechanism. Last meeting, we allocated up to and including 1055. Alex has a paper explaining RCS. He distributed a copy to all members in attendance. IR Updating Discussion (all) ---------------------------- The group discussed the need for additional rules for superseding the old IRs. After discussion, it was decided that: 1. Superseded old IRs will have the "1993 Disposition," "Disposition Rationale," and "Superseded By" fields inserted with the appropriate contents. Nothing else will be changed. 2. Closed old IRs should be treated similarly. 3. When creating a new IR from and old one, don't change the author field unless major updates are made, in which chase, the superseder's name should be used. 4. For new IRs, do the new fields "1993 Disposition," "Disposition Rationale," and "Superseded By" make any sense? The consensus (lukewarm) is that these fields should be retained and given the value "N/A." Typo IR (Clive) --------------- We will open a "Typo IR," in which we will collect all typos. This IR will be IR 1000. Daniel Barclay will be the owner. IR Voting Mechanism (all) -------------------------- There was discussion on the need for a more formalized mechanism for approval of IRs within the ISAC. This discussion gradually evolved into a discussion of how to coordinate between 1076 and the other groups building on top of 1076. It was resolved that the ISAC chair (Clive) will invite the chairs of DASC SGs and WGs to participate in the ISAC mailing list, so that they can receive early information about changes of potential interest to them. The current DASC Steering Committee Members are: Paul Menchini DASC Chair mench@mercury.interpath.net Jean Mermet DASC Vice Chair mermet@imag.fr Victor Berman DASC Treasurer berman@cadence.com John Willis DASC Secretary willis@vnet.ibm.com Ron Waxman U.S. TAG Representative ronw@virginia.edu John Hines Ex-Officio hines@el.wpafb.af.mil Clive Charlwood VASG WG Chair crc@synopsys.com Steve Bailey VHDL Shared Variable WG Chair sab@vas.com Jean-Michel Berge VHDL Analog Extensions WG Chair berge@cns.cnet.fr Jose Torres VHDL Math Package WG Chair jose@synopsys.com Alex Zamfirescu VHDL Synthesis Package WG Chair a.zamfirescu@ieee.org Victor Berman VHDL Timing Methodology WG Chair berman@cadence.com Gabe Moretti VHDL Utility Library WG Chair gabe@dazixca.ingr.com J. Bhasker VHDL/EDIF Interoperability WG Chair jb7@mhcnet.att.com Maq Mannan Verilog ASG WG Chair mannan@galaxy.nsc.com John Hillawi At-Large hillawi@cix.compulink.co.uk Matt Imai At-Large imai@tutics.tut.ac.jp@relay.cs.net The Study Group Chairs are: Dave Barton SDDL SG Chair dlb@wash.inmet.com Doug Dunlop OOVHDL SG Chair dunlop@wash.inmet.com Zain Navabi VHDL Test SG Chair navabi@nuvlsi.coe.neu.edu John Willis VHDL Parallel Simulation SG Chair willis@vnet.ibm.com Database Issues (Alex) ---------------------- A general discussion of protocols, procedures, and mechanisms for updating and distributing the IR database ensued. There also is a need to have written procedures for installing and updating IRs. Alan Dewey volunteered to take this on. Alex will assist. What we have: IRs stored under RCS. Individual login ids for all ISAC members, with appropriate privileges. LRM(s). Paul will ensure that a Mac version is installed. What is needed: Allocate new IR number, RCS checkout w/lock, RCS checkout w/o lock, RCS check-in, create new IR, release lock. All facilities should be email accessible (as well as telnet accessible), and all check-ins should broadcast the IR. We also need written protocols for the use of these services. Disposition of "Unknown" IRs ---------------------------- 6: Alex owns this IR. Superseded to IR 1074. 41: Jacques owns. Closed. Section 10.3 addresses the issue of visibility of items in configuration declarations. 48: Alex owns. It will be closed, but a subsidiary issue will be developed as IR 1001. 59: Assigned to Victor. Closed. Section 7.2 contains the suggested modifications. 64: Assigned to Paul. Closed. Sections 5.1 and 10.3 contain the necessary language. 67: Has to do with the staticness of names. Closed. A new IR (1056) will be opened. Daniel will do. 77: Assigned to Andy. Superseded to IR 1064. 78: Assigned to Alan. Closed. Section 8.8 contains the necessary language. 80: Chuck reports that this issue is closed. The issue is fundamentally addressed in Section 7.4.2, although there may be new issues related to staticness. Any such new issues will be the focus of a new IR or IRs, as is appropriate. 81: Bill reports that this issue is closed. Section 7.4 contains the necessary language. 85: Assigned to Bill. Closed. Section 5.2.1.2 contains the necessary language. 86: Assigned to Chuck. Closed. The IR does not modify the LRM. However, the introduction of impure functions may introduce new but related issues. A new IR 1065 should address these issues. Paul will write. (Among the issues are order of association, how many times expressions are evaluated in generic maps of component specifications of the form "for all" or "for others", etc.) 95: Closed by Daniel. Section 12.6.2 completely addresses this issue. 100: Assigned to Alex. Transmitted to non-ISAC. 113: Supersede to IR 1057. Daniel owns. 122: Transmitted to non-ISAC. Daniel owns. 127: Assigned to Daniel. Superseded to 1066. 140: Partly but incompletely addressed. Superseded by 1058. Daniel owns. 145: Assigned to Jacques. Recursive configuration employing configuration specifications is clearly allowed; however, it is unclear whether configuration declarations can be used to describe recursive (and mutually recursive) configurations. Superseded to 1067. 154: Victor owns. Superseded to 1075. 158: Not addressed in 1076-1993. Superseded by 1059. May be other issues about legality of others in aggregates, etc. Daniel owns. Paul will find and send his previous work on this area to Daniel. 160: Oz owns. No report as yet. Given to Jacques for analysis and disposition. 163: Assigned to Clive. Superseded to 1068. 171: Assigned to Victor. Closed. Sections 7.4.1 (and possibly 7.2) have addressed the issue. 172: Assigned to Paul. Superseded to 1069. 176: Chuck has superseded this as 1006. 181: Assigned to Andy. Closed. The LRM clearly states that formal ports may be unassociated under certain circumstances. 190: Assigned to Alan. This IR touches on the issue of outer scope and the nature of library names. Superseded to 1070. 194: Closed by Daniel. Chapter 0 addresses. 203: Daniel owns. Not addressed. Section 3.1 addresses type of a range. Daniel will supersede to 1060. Daniel will open a new IR (1061) on the type of a range. 207: Assigned to Bill. Superseded to 1071. 208: Assigned to Chuck. Closed. Language correcting the problem has been added to 4.3.1.1. 217: Assigned to Alex. Superseded to 1072. 219: Chuck and Paul own. Superseded to 1076. Chuck to write up. 223: Oz owns. No report as yet. Given to Andy and Alan for analysis and disposition. Superseded to Daniel's IR 1057. Andy will update 223. 226: Assigned to Daniel. Closed. Section 4.3.2.1 contains language describing the meaning of default expressions on interface objects. 229: Bill owns. Closed. Section 5.1 contains the necessary language. 235: Assigned to Jacques. Closed. The subtype of the result of concatenation is now well-defined. 239: Daniel will supersede to IR 1062. 240: Assigned to Clive. Closed. Section 2.1.1.2 clears this up. 243: Assigned to Victor. Superseded to IR 1063. 244: Assigned to Paul. Superseded to 1073. 253: Assigned to Andy. Closed. Section 13.7 allows bit string literals to be of any character array type whose element subtype contains '0' and '1'. They may also be of zero length. Disposition of 1993 Comments Forwarded to ISAC ---------------------------------------------- Many issues were forwarded to the ISAC during 1076-1993 ballot resolution. These must be addressed. But before these can be addressed, we need to stabilize the state of the old IRs. This stabilization will be addressed first. However, in the meantime, Paul will send Alex the ballot resolution documents so that Alex can put them in the repository. The ISAC unanimously agreed to first attack IRs that are considered important, then to distill the 1993 comments into new IRs, then (finally) any remaining IRs. The intent is to assure that the coming 1993 implementations work the same. IR Status --------- The present IR status is as follows. Clive will put this list on-line in the old IR database. 1 Closed by Paul (Section 7.2.1 line 77 in C+ LRM) 2 Closed by Chuck 3 Closed by Oz 4 Non-ISAC issue 5 Closed by Alex 6 Alex owns this IR. Superseded to IR 1074. 7 Closed by Victor 8 Superseded by IR1004 (John will write) [Note, the last minutes stated "Superseded by IR0004....-Ed.] 9 Closed by Bill 10 Closed by Paul (There are several null characters in file) 11 Superseded by IR1003 (Chuck will write) 12 Closed by Oz 13 Closed by Clive 14 Closed by Alex 15 Closed by Daniel 16 Closed by Victor 17 Closed by John 18 Closed by Bill 19 Closed by Paul (Section 10.3 was revised) 20 Transmitted to non-ISAC resolution (Chuck) 21 Closed by Oz 22 Closed by Clive 23 Closed by Alex 24 Superseded by IR1008 (Daniel will write) 25 Closed by Victor 26 Closed by John 27 Closed by Bill 28 Closed by Paul (Section 5.2.1.2 was revised) 29 Transmitted to non-ISAC resolution (Chuck) 30 Closed by Oz 31 Closed by Clive 32 Closed by Alex 33 Closed by Daniel 34 Closed by Victor 35 Superseded by IR1011 (John will write) 36 Closed by Bill 37 Closed by Paul, superseded by IR0063 38 Closed by Chuck 39 Closed by Oz 40 Closed by Clive and Paul 41 Jacques owns. Closed. Section 10.3 addresses the issue of visibility of items in configuration declarations. 42 Closed by Daniel 43 Closed by Victor 44 Closed by John 45 Closed by Bill 46 Closed by Paul (Section 2.4 line 383 was revised, Section 10.4) 47 Closed by Chuck 48 Superseded by IR1001 (Oz, Dan, Alex will write) 49 Closed by Clive 50 Closed by Alex 51 Closed by Daniel 52 Closed by Victor 53 Superseded by IR1012 (John will write) 54 Closed by Bill 55 Superseded by IR1017 (Paul will write) 56 Superseded by IR1005 (Chuck will write) 57 Closed by Oz 58 Closed by Clive 59 Assigned to Victor. Closed. Section 7.2 contains the suggested modifications. 60 Closed by Daniel 61 Closed by Victor 62 Closed by John 63 Closed by Bill 64 Assigned to Paul. Closed. Sections 5.1 and 10.3 contain the necessary language. 65 Closed by Chuck (Chuck will write IR1006 on wording) 66 Closed by Oz (Alex may re-examine and open another issue) 67 Has to do with the staticness of names. Closed. A new IR (1056) will be opened. Daniel will do. 68 Closed by Alex 69 Closed by Daniel 70 Closed by Victor 71 Closed by John 72 Superseded by IR1013 (Bill will write) 73 Closed by Paul (Now have declarative region for generates) 74 Closed by Chuck 75 Superseded by IR1002 (Oz will write) 76 Closed by Clive 77 Assigned to Andy. Superseded to IR 1064. 78 Assigned to Alan. Closed. Section 8.8 contains the necessary language. 79 Closed by Victor, Alex dissented and may open a new issue 80 Chuck reports that this issue is closed. The issue is fundamentally addressed in Section 7.4.2, although there may be new issues related to staticness. Any such new issues will be the focus of a new IR or IRs, as is appropriate. 81 Bill reports that this issue is closed. Section 7.4 contains the necessary language. 82 Closed by Paul (Section 7.4 addresses this) 83 Superseded by IR1007 (Chuck will write) 84 Closed by Oz 85 Assigned to Bill. Closed. Section 5.2.1.2 contains the necessary language. 86 Assigned to Chuck. Closed. The IR does not modify the LRM. However, the introduction of impure functions may introduce new but related issues. A new IR 1065 should address these issues. Paul will write. (Among the issues are order of association, how many times expressions are evaluated in generic maps of component specifications of the form "for all" or "for others", etc.) 87 Superseded by IR1009 (Daniel will write) 88 Superseded by IR1010 (Victor will write) 89 Closed by John 90 Closed by Bill 91 Closed by Victor 92 Superseded by Paul to IR0242 93 Closed by Chuck 94 Closed by John 95 Closed by Daniel. Section 12.6.2 completely addresses this issue. 96 Closed by John 97 Superseded by IR1018 (Oz will write) 98 Closed by Clive 99 Closed by Andy 100 Assigned to Alex. Transmitted to non-ISAC. 101 Closed by Paul (Section 7.3.5 revised) 102 Superseded by IR1014 (Chuck will write) 103 Closed by Bill 104 Transmitted to non-ISAC issue 105 Superseded by IR1015 (John will write) 106 Superseded by IR1019 (Oz will write) 107 Superseded by IR1016 (Clive will write) 108 Closed by Andy 109 Closed by Victor 110 Transmitted to non-ISAC issue 111 Transmitted to non-ISAC issue 112 Closed by Bill 113 Supersede to IR 1057. Daniel owns. 114 Closed by John 115 Closed by Oz 116 Closed by Clive 117 Closed by Andy 118 Closed by Victor 119 Closed by Paul (Section 9.5 line 241 was revised) 120 Closed by Chuck 121 Closed by Bill 122 Transmitted to non-ISAC. Daniel owns. 123 Transmitted to non-ISAC issue 124 Closed by Oz 125 Closed by Clive (Section 5.3 was revised) 126 Transmitted to non-ISAC issue 127 Assigned to Daniel. Superseded to 1066. 128 Superseded by IR1026 (Paul will write) 129 Closed by Chuck (Section 10.3 was revised) 130 Closed by Bill (Section 4.3.2.2 was revised) 131 Closed by Daniel (Superseded by IR0048) 132 Superseded by IR1027 (John will write) 133 Transmitted to non-ISAC issue 134 Transmitted to non-ISAC issue 135 Superseded by IR1028 (Andy will write) 136 Superseded by IR1017 (Victor will write) 137 Transmitted to non-ISAC issue 138 Transmitted to non-ISAC issue 139 Closed by Bill (Section 6.6, 14.1 revised) 140 Partly but incompletely addressed. Superseded by 1058. Daniel owns. 141 Superseded by IR1029 (John will write) 142 Superseded by IR1020 (Oz will write) 143 Closed by Clive 144 Transmitted to non-ISAC issue 145 Assigned to Jacques. Recursive configuration employing configuration specifications is clearly allowed; however, it is unclear whether configuration declarations can be used to describe recursive (and mutually recursive) configurations. Superseded to 1067. 146 Closed by Paul (Section 4.2 and Section 10.5 was revised) 147 Transmitted to non-ISAC issue 148 Transmitted to non-ISAC issue 149 Superseded by IR1030 (Daniel will write) 150 Closed by John (Section 5.2.1.2 was revised) 151 Superseded by IR1021 (Oz will write) 152 Transmitted to non-ISAC issue 153 Superseded by IR1031 (Andy will write) 154 Victor owns. Superseded to 1075. 155 Superseded by IR1023 (Oz owns this, note multiple references) 156 Superseded by IR1032 (Chuck owns this) 157 Superseded by IR1033 (Bill will write this) 158 Not addressed in 1076-1993. Superseded by 1059. May be other issues about legality of others in aggregates, etc. Daniel owns. Paul will find and send his previous work on this area to Daniel. 159 Superseded by IR0182 160 Oz owns. No report as yet. Given to Jacques for analysis and disposition. 161 Closed by Clive (Section 12.1 revised this) 162 Superseded by IR1034 (Andy will write) 163 Assigned to Clive. Superseded to 1068. 164 Closed by Paul (Section 7.3.2 was revised) 165 Closed by Chuck (Section 13.7 was revised) 166 Closed by Bill 167 Transmitted to non-ISAC 168 Closed by John (Section 7.2.4 was revised) 169 Superseded by IR1022 (Oz will write) 170 Closed by Clive 171 Assigned to Victor. Closed. Sections 7.4.1 (and possibly 7.2) have addressed the issue. 172 Assigned to Paul. Superseded to 1069. 173 Superseded by IR1035 (Paul will write) 174 Closed by Chuck (Section 14.1) 175 Transmitted to non-ISAC 176 Chuck has superseded this as 1006. 177 Closed by John (See IR0094) 178 Superseded by IR1023 (Oz will write) 179 Closed by Clive (Manual was misinterpreted) 180 Closed by Andy (Manual was misinterpreted) 181 Assigned to Andy. Closed. The LRM clearly states that formal ports may be unassociated under certain circumstances. 182 Transmitted to non-ISAC issue 183 Closed by Chuck (Section 4.3.1.4 was revised) 184 Superseded by IR1012 (note multiple references) 185 Closed by Daniel (already superseded by IR0026) 186 Superseded by IR1047 (John will write) 187 Transmitted to non-ISAC by Oz 188 Closed by Clive (previously superseded by IR0050) 189 Superseded by IR1036 (Andy write this) 190 Assigned to Alan. This IR touches on the issue of outer scope and the nature of library names. Superseded to 1070. 191 Closed by Paul (Previously superseded by IR0178) 192 Closed by Chuck (Section 5.1 was revised) 193 Closed by Bill (Section 3.2.1 discusses this) 194 Closed by Daniel. Chapter 0 addresses. 195 Closed by John (Superseded by IR0113) 196 Superseded by IR1024 (Alex will write) 197 Closed by Clive (Manual was misinterpreted) 198 Closed by Andy (Manual was misinterpreted) 199 Closed by Victor 200 Transmitted as a non-ISAC issue 201 Transmitted as a non-ISAC issue 202 Superseded by IR1038 (Bill will write) 203 Daniel owns. Not addressed. Section 3.1 addresses type of a range. Daniel will supersede to 1060. Daniel will open a new IR (1061) on the type of a range. 204 Closed by John (Section 8.1 was revised) 205 Closed by Oz (Section 4.3... was revised) 206 Transmitted as a non-ISAC issue 207 Assigned to Bill. Superseded to 1071. 208 Assigned to Chuck. Closed. Language correcting the problem has been added to 4.3.1.1. 209 Superseded by IR1039 (Paul will write) 210 Transmitted as a non-ISAC issue 211 Transmitted as a non-ISAC issue 212 Transmitted as a non-ISAC issue 213 Closed by John (Section 12.1 was revised) 214 Closed by Oz (Section 12.6.4 was revised) 215 Closed by Clive (Section 11.2 was revised) 216 Superseded by IR1040 (Andy will write) 217 Assigned to Alex. Superseded to 1072. 218 Superseded by IR1041 (Paul will write) 219 Chuck and Paul own. Superseded to 1076. Chuck to write up. 220 Closed by Bill (Section 7.4 was revised) 221 Superseded by IR1042 (Daniel will write) 222 Superseded by IR1043 (John will write) 223 Oz owns. No report as yet. Given to Andy and Alan for analysis and disposition. Superseded to Daniel's IR 1057. Andy will update 223. 224 Superseded by IR1044 (Clive will write) 225 Superseded by IR1045 (Andy will write) 226 Assigned to Daniel. Closed. Section 4.3.2.1 contains language describing the meaning of default expressions on interface objects. 227 Closed by Paul (Section 11.4, paragraph was removed) 228 Closed by Chuck (Section 9.6 was revised) 229 Bill owns. Closed. Section 5.1 contains the necessary language. 230 Closed by Daniel and Paul 231 Closed by John (Note 1 of Section 5.2.1.2 addresses this) 232 Transmitted to non-ISAC status by Oz 233 Superseded by IR1046 (Chuck will write) 234 Superseded by IR1047 (John will write) 235 Assigned to Jacques. Closed. The subtype of the result of concatenation is now well-defined. 236 Superseded by IR1048 (Paul will write) 237 Superseded by IR1049 (Chuck will write) 238 Closed by Bill 239 Daniel will supersede to IR 1062. 240 Assigned to Clive. Closed. Section 2.1.1.2 clears this up. 241 Superseded by IR0156 (Chuck will pursue) 242 Superseded by IR1050 (Clive will write) 243 Assigned to Victor. Superseded to IR 1063. 244 Assigned to Paul. Superseded to 1073. 245 Superseded by IR1051 (Paul will write) 246 Superseded by IR1046 (Chuck will write) 247 Superseded by IR1052 (Bill will write) 248 Superseded by IR1053 (Daniel will write - This is a big one) 249 Superseded by IR1050 (John will write) 250 Superseded by IR1025 (Oz will write) 251 Closed by Clive (Section 4.1.1 bullet 9 addresses this) 252 Superseded by IR1054 (Andy will write) 253 Assigned to Andy. Closed. Section 13.7 allows bit string literals to be of any character array type whose element subtype contains '0' and '1'. They may also be of zero length. 254 Superseded by IR1055 (Paul will write) New IRs ------- Numbers were first assigned to some new IRs: 1077 Out ports cannot drive buffer ports. 1078 Memory leak and portability issue with TextIO 1079 Conflicting rules for direction of concatenation 1080 Implicit alias when alias type 1081 Expression elaboration with impure functions 1082 Implicit conversion of a universal real 1083 Pureness of EndFile 1084 Use of keyword OPEN in incremental binding 1085 Staticness of the predefined attribute Instance_Name 1086 Use of character literal or operator symbol in alias 1087 Subtype checks when port and signal have different types These were next sorted by importance into two categories: Most important: 1079, 1080, 1083, 1084, 1085. Of lesser importance: 1077, 1078, 1081, 1082, 1087. More information needed: 1086. IR 1085 ------- The author appears to be correct: PATH_NAME is never locally static, while (through an omission), INSTANCE_NAME appears to be. Paul will write up and distribute--as part of this write-up, all instances of the attributes will be searched for to ensure consistency. It will be considered at the May meeting; if consensus is reached, it will be presented to the VASG at the DAC meeting. IR 1080 ------- Once again, the author appears to be correct; the author of the ISAC write-up should ensure that if, as the issue author suggest, we substitute "predefined subprograms" instead of "predefined operators," that "subprograms" includes "operators"--language in Section 7.4.1, Locally Static Primaries, is germane. Victor will write up. During the discussion, several related issues surfaced: e.g., the relationship between use clauses and aliases of the used declarations. Alex will research these topics; Chuck will solicit Model Technology's perspectives, Leda will also provide their perspective. The results will be discussed at the May meeting. IR 1079 ------- There indeed is an inconsistency in the LRM related to some old text that was not modified/deleted when the LRS on concatenation was implemented. The general feeling is that removal of the first definition presented in the IR is a sufficient change; however, if this definition is retained, not only must the second definition be modified/deleted, but other, unspecified, changes must also be made. Alan will write this up and distribute. IR 1084 ------- The consensus of the group is that the example in the proposed resolution appears to be partially in error. According to Section 5.2.1, bullet 4, the incremental rebinding of the port P1 is in error. However, there is some disagreement as to the meaning of the rebinding of the generic N. It appears that the second bullet of Section 5.2.1 applies, but this paragraph does not explicitly state the meaning of OPEN in an incremental generic map. Therefore, there appear to be two interpretations possible: 1. An actual of OPEN means exactly the same as no association in the rebinding. Under this interpretation, the value of N is 8. 2. The actual of OPEN "undoes" the primary binding. Under this interpretation, the value of N is 5. There were questions as to the value or danger (from a human-engineering standpoint) of allowing OPEN in an incremental generic map to undo the effect of the primary binding. Hence, there were some questions as to whether rebinding with OPEN should be legal. It was noted that rebinding with OPEN was never considered during the 1993 restandardization. Thus, there are three possibilities: The first two are the two interpretations noted above; the third is to make rebinding with OPEN illegal. A subsidiary question is: "What is the value of N if both the primary and incremental bindings of N are OPEN?" There seems to be consensus that the value of N is 5 in this circumstance. This question was posed in an effort to provide unification, but did not appear to provide any insight. Since no consensus was reached, and other related questions about OPEN were raised, Daniel agreed to analyze the IR, come up with one or more proposed resolutions, and present at the May meeting. 1083 ---- There appears to be consensus that: 1. EndFile is an impure function, and 2. that pure functions should not have access to files. There was some debate about how to best implement the second item of consensus. The two choices are: 1. Add the restriction that pure functions may not have file parameters. 2. Disallow the occurrence of file operations from within the bodies of pure functions. It was observed that approach 1 does not require the analysis of the subprogram body to detect the error, while approach 2 requires analysis of the function's body. Based on this observation, consensus was quickly reached on the first approach. A subsidiary question was: "Can a pure function declare a file object within it's body?" Although it appears that it can, it is clear that it cannot be referenced and therefore cannot be used. It seems worth disallowing this case for pure functions, since a pure function could otherwise be written that can query the existence of external files. It also appears that there is another potential holes. A pure function declared within an impure function could reference a file parameter of the impure function. This case should also be disallowed. There was some discussion as to whether asserts should be disallowed in pure functions. After some discussion, the consensus seemed to be that they should continue to be allowed. Andy will write this up. 1003 ---- This IR has been worked on sporadically for two years. There may be some terminological issues, but it appears to be largely done. If so, should we complete it? Some other issues are superposition invariance and reflexivity, and whether attributes should be supplied that can be used to notify a simulator that a given resolution function has or does not have these properties. The group noted that there is an entry in the "Non-portable constructs" appendix about such resolution functions. A vote was taken on the IR as it now stands, with the result that the IR is ISAC-Approved. It will be presented at the DAC by Chuck. Next Meetings ------------- 5 May, at the Claremont Hotel in Oakland California, in conjunction with VIUF. Alan will make the local arrangements. 28-30 July, Ashville, North Carolina. Please RSVP to Paul Menchini (mench@mercury.interpath.net) by 5 May. Chair's Closing Comments ------------------------ Clive appealed to all ISAC members to get their work done outside of the meeting. He asks all members to review these minutes and complete all existing assignments prior to the May meeting. Respectfully submitted, Paul Menchini