ISAC: Minutes from meeting on 07 December 2006

From: Chuck Swart <cswart_at_.....>
Date: Thu Dec 07 2006 - 21:24:59 PST
Attached are the minutes and also the summary status document for
pre-2002 IRs.


Last edited 02 September 2005

Analysis of VHDL93 IRs for VHDL2002

1001    ResolvedEntity aspect in default binding indication is unclear.
1002    Open     Ambiguity as to when parameter subtype indications are elaborated.
1003    Open     Non-commutative and non-associative resolution functions are a source of non-determinism
1004    Enhancement-Analysis needed Are deferred constants still deferred after their full declaration?
1005    Open     Unclear requirements on prefix in predefined attribute names.
1006    Resolved Parentage checks are difficult to verify during static model elaboration.
1007    Resolved Can a range constraint in a type declaration use 'RANGE?
1008    Resolved Function NOW is undefined during static elaboration.
1009    Open     Unclear context of evaluation for formal part of block map aspects.
1010    Open     Overloaded convertible operands make type analysis very difficult.
1011    Open     Enumeration literals, physical units, and library names cannot be attributed.
1012    Open     It is unclear how Physical Literals of type TIME should be treated when checking for conformance.
1013    Open     Reference to a generate loop parameter in a configuration is unclear
1014    Open     The LENGTH attribute is ill-defined.
1015    Open     The note illustrating the equivalent sequence of declarations for a constrained array type definition is
1017    Open     Why must a name in a wait stmt. sens. list be static?
1018    Open     What does "reference" mean?
1019    Resolved The signal updating steps of the simulation cycle is incorrect.
1020	Open	 The 'note' implies that predefined attributes of the actual will not be visible on the formal.
1021	Open	 Constrained arrays with negative index bounds are difficult to declare.
1023    Open     VPI Issue 2 -- Specifications applying to 0 entities
1024    Open     VPI Issue 20 -- Primary units with same name
1026    Resolved Glossary refers to old "size constraint" in entry for "constraint."
1027    Resolved  REVERSERANGE should be a keyword in the language.
1028    Open     Is the entity identifier directly visible in architecture bodies and configuration declarations?
1029    Reopen   What should happen when there is a port clause in a block statement but no port map clause?
1033    Open     Attributes cannot be specified in package bodies.
1034    Open     Qualified expression checks may be too restrictive.
1038    Open     The definitions of A'LEFT, A'RIGHT, A'HIGH, A'LOW, A'RANGE, and A'REVERSE_RANGE are ill-defined when
1039    Open     May subelements and slices of procedure signal parameters be individually driven?
1044    Resolved Definition of 'HIGH and 'LOW in a null range
1053    Resolved This IR number (1053) was allocated but not needed.
1054    Reopen   Short-circuit operators don't when called as functions
1068    Open     When do buffer ports have sources?
1073    Resolved The meaning of a qualified expression is in dispute
1078    Open     Memory leak and portability issue with TEXTIO
1083    Open     Pureness of function ENDFILE

Still to be analyzed:
1000 Accumulated typographical and terminology errors. 
1016 LRM definition of block statements is ambiguous.
1022 A subtype Delay would aid VHDL definition and use.
1025 Checking the target type of an aggregate signal assignment
1030 Imprecise wording and inconsistent presentation order.
1031 Can targets be in the form of nested aggregates?
1032 May analyzers make non-locally static checks?
1033 Attributes cannot be specified in package bodies.
1035 Rules for non-local access to signals and variables in functions are too restrictive.
1036 Integer subtypes & operations
1040 Resolved composite signals are difficult to associate.
1041 Implicit signals are not objects.  Moreover, they may not be signals.
1042 Missing sentence for variable assignment aggregate target.
1043 User defined attribute on function name
1045 Qualified expressions lead to lexical ambiguity
1046 Unclear definition of "resolved signal."
1047 VPI Issue 10 -- Entity class type vs. subtype
1048 Deferred constant reference before complete definition.
1049 When is an entity bound to the library it references?
1050 Visibility outside of design units is not well-defined.
1051 Direct visibility rules for USE clauses are incomplete and confusing.
1052 Rules for others in aggregates are incomplete
1055 Reading predefined signal attributes of subprogram signal parameters within the subprograms.
1056 Certain error requirements are not well-defined.
1057 Expanded names are not allowed in extended regions.
1058 What use are LINKAGE ports?
1059 What is the allowed use of others in record aggregates?
1060 A'LEFT's result type is anomalously described.
1062 Can a resolution function return an unconstrained array?
1063 Scope and visibility rules for predefined attributes are unclear.
1064 Passiveness of processes is not analysis-time computable.
1066 Conversion function definition assumes different types.
1069 Mod and rem are ill-defined when they return zero.
1070 VPI Issue 14 -- Prefixes in USE clauses
1071 The bounds of array aggregates are ill-defined.
1072 Does the elaboration of a process statement create a process?
1074 Interpretation of overloaded name prefixes is unclear.
1075 Self-referential access types are not prohibited.
1076 Applicability of resolution functions is still unclear.
1077 Out ports cannot drive buffer ports.
1079 Conflicting rules for direction of concatenation
1081 Expression elaboration with impure functions
1082 Implicit conversion of a universal real expression
1085 "Staticness" of the predefined attribute INSTANCE_NAME
1086 Use of character literal or operator symbol in alias
1088 Problems with physical mapping of libraries
1089 Visibility problem with components and USE clauses
1090 Do library clauses extend from entities to architectures?
1091 Group constituent names cannot have signatures.
1092 Must a selected name in a use clause denote anything?
1093 Replacement characters (from Ada) complicate parsing.
1094 Bad wording for individual association for subprograms.
1095 Are time literals with real numeric parts exact?
1096 Entity aspect poorly defined for incremental binding.
1097 Why do primary unit names in secondary units differ?
1098 Can variable parameters be left unassociated or not?
1104 Inconsistent, overly restrictive subtype checks.
1105 The elaboration of group declarations is not defined.
1106 Accumulated problems with unclear wording.
1107 Postponed process behave illogically at initialization.
1108 The subtype of a deferred constant may not always be deduced during analysis time.
1109 The term "named entity" it not defined.
1110 The term "object" is not well-defined.
1111 Is a user-defined attribute an object?
1112 Need to disallow conversion function signal parameters.
1114 LRM wording doesn't account for separators.
1115 Should subtype and constraint checks really be made when the driver takes on the value of a signal?
1116 implicit signals in guard expression (via port maps)
1117 Problem with Composite Resolved Signal
1118 Group constituent can't be user-defined attribute name.
1119 LRM is not clear as to when the wait statement condition clause is evaluated in a postponed process.
1120 Inconsistency about the mode of formal parameters of class file.
1121 Meaning of a changed value for reals is not clear
1122 Default binding text and examples disagree
1123 Simulation cycle does not account for processes resuming after expiration of the timeout interval
1124 Are all actuals evaluated in incremental binding?

Minutes of ISAC meeting held via telecom on 07 December 2006

Present: Peter Ashenden, Jim Lewis, Larry Soule, Chuck Swart, Ajay
Verikat

Absent: Lance Thompson

Next Meeting: Thursday, January 4, 2007, 6 pm Pacific Time
              (Friday, January 5, 2007, 2 am GMT )

TOPIC: Bugzilla.

We discussed the possibility of moving the IR system to Bugzilla.
One big advantage is that we would have an automatic tracking
capability, so the odds of losing a submission, change, comment
would be very low. Chuck's only major concern is that IRs with
complicated analyses would be very hard to follow unless submitted
text could be edited. Also it would be good to be able to perform
minor routine edits (spell checking, formatting paragraphs, etc)
Complicated analyses could be put in attachments, but its not clear
if any editing can be done on the main entries.

It was also observed that it would be nice if someone could add
his name to receive email for any bug on a give project.

Peter has suggested a reorganization of the current Bugzilla system.
Chuck will forward his suggestion to ISAC and to the Accellera LRM
group for comment.

It was tentatively agreed that we would begin using Bugzilla for
ISAC issues starting with the next IEEE release.

ACTION: Chuck to pursue editing details

TOPIC: IR 2105 Can't declare an alias of a character literal without
using expanded name.

This looks like a language oversight.

ACTION: Chuck to analyze.

TOPIC: IR 2099 Alias declarations introduce homographs

Peter supplied an email which commented on Chuck's latest revision.
Peter's example shows that the real issue is not restricted to the
"identity aliases" in Chuck's analysis. The more general idea is that
two nonobject aliases with the same alias designator are not homographs
if they refer to the same named entity. (Similarly, an alias and an
identifier, character_literal, operator_symbol are not homographs
if the named entity in the alias declaration is  that identifier.,,,)
In both these cases, for overload resolution, the two aliases, or the
alias and the declaration are considered to be multiple references to
the same named entity. Peter also pointed out a minor omission in the
definition of "identity alias": the alias designator is the same as the
simple name, character literal, or operator symbol of the denoted name.

ACTION: Chuck to continue analysis, then all to review.

TOPIC: Review of IRs incorporated into D3.0

2044   Ajay   OK
2045   Chuck  OK
2047   Jim    OK (IR issued on typo)

ACTION: IRs to be examined for next meeting:

2043    015 Larry
2048    019 Ajay
2049    020 Ajay
2050    021 Chuck
2051    022 Chuck
2052    022 Chuck
2053    023 Larry
2055    025 Larry
2056    026 Lance
2057    027 lance

TOPIC: Review of Pre-2002 IRs

1015   Chuck  Open issues
1016   Larry
1018   Ajay   Open issues
1020   Chuck  Open issues
1021   Lance  Open issues

ACTION: Pre-2002 IRs to be examined for next meeting:

1000    Peter    ongoing
1022    Ajay
1025    Ajay
1026    Larry    (All issues resolved)
1027    Larry    (All issues resolved)
1030    Chuck
1031    Chuck
1032    Lance
1033    Lance
Received on Thu Dec 7 21:25:03 2006

This archive was generated by hypermail 2.1.8 : Thu Dec 07 2006 - 21:25:04 PST