[sv-champions] Meeting minutes of August 8 - date now fixed

From: Neil Korpusik <Neil.Korpusik_at_.....>
Date: Sat Aug 18 2007 - 19:28:21 PDT
There was a typo on the date shown in the minutes.
It is now fixed.

Neil




-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.


Champions meeting minutes of August 8, 2007
8am - 10am PST

1. A  Dave Rich
2. A  Neil Korpusik
3. A  Surrendra Dudani
4. A  Brad Pierce
5. A  Shalom Bresticker
6. A  Francoise Martinolle 
7. -  Stu

Note: Stu was in Japan during our meeting. He sent input to the committee
      ahead of time mentioning that he had reviewed all of the mantis items
      from the Editor's standpoint and did not have problems with any of them.


1. Review IEEE patent policy
   http://standards.ieee.org/board/pat/pat-slideset.ppt

      Move: Dave - assume that the patent policy was read
    Second: FM

2. Mantis items

From the sv-ac

 1674   Context value functions
	$inferred_clock
	$inferred_disable
	$inferred_enable
    approved unanimously by email vote 06/25/07

    Stu  - ok
    Dave - I am opposed to using system functions that represent syntactic 
	   sugar. I think syntax is better at representing syntactic sugar. For 
	   example "clk = default clocking"
         - The SV-AC is adding a lot of system functions.
	 - Seem to not be putting enough thought into language space solutions.
	 - Not against the feature itself.
  Shalom - Dmitry sent email addressing Dave's feedback.
	   "System functions are widely used in SystemVerilog
	   for this purpose. $typename is just one example."
    Dave - They are only usable in a property, and only as default argument of 
	   a property.
	 - Has suggested the possibility of using a method. 
    FM   - Not sure that a method would be a good approach
	   Maybe ok if associated with a property.
  Shalom - Restricting enhancements to their part of the language.
   Dave  - This one is fairly isolated - doesn't feel too strongly about it
	 - This is just one example of their use of system functions
	   properties will end up looking like a bunch of system functions
         - If PLI is to override these tasks, how would it access that info?
     FM  - PLI can usually override them, an advantage of system functions?
    Dave - Not using a system function because they need it, using it more like 
	   a language construct. Only meaningful in a particular position. 
	   Should be using grammar for this. 
  Shalom - Can avoid new keywords this way.
	   Is there really a problem with adding new system functions?
	 - They are thinking about it as a query. 
         - Elaboration time system functions
    Dave - $typename is another one
	 - The sv-ec would typically use a method or the :: syntax
	 - A method of the property would be possible.
    Brad - This proposal is similar to $bits
  Shalom - These are context dependent
	 - more like 'this'
   Sur   - could be part of a context object
   FM    - Thinks system function is ok
         - It is unfortunate that they are only for properties.
  Shalom - sv-ac thought it would be easier if they restricted it to properties
   Dave  - ## can refer to a clocking event.
  Shalom - That is an event control
	 - These are for binding purposes, explicit or inferred
    Dave - withdrew his objection

        Move: Brad - approve the proposal for mantis 1674
      Second: Surrendra
      Passed unanimously


 1677   Add $changed sampled value function
    approved unanimously - 02/06/07 meeting

    Shalom - Just adding another system function to an existing set of system 
	     functions.
    Brad   - It is the opposite of $stable.

        Move - Francoise - approve the proposal for mantis 1677
      Second - Shalom
      Passed unanimously


 1704   need to specify behavior of attached subroutine on empty seq match
    approved unanimously by email vote 03/01/07

     Brad  - "must not admit" should be  
    Shalom - "shall not admit"
     Sur   - a compiler could check this
	   - drop commas in the 2nd to last sentence
     Dave  - what does admit mean?  
     Sur   - not well defined in the LRM
	   - means "cannot have an empty match"
     Brad  - that word is already in the LRM
           - need to get a compiler error
	   - "Shall be a compile time error..."
     Dave  - let the committee figure out the exact language required.

	Three changes need to be made:
	   - "must not admit" should be "shall not admit"
	   - Get rid of commas in the next to last sentence.
	   - When used on a sequence that admits an empty match an error 
	     message shall be issued. [The sv-ac should determine the 
	     appropriate wording for this error case.]

        Move: Dave - send the proposal for mantis 1704 back to svac (3 issues)
      Second: Brad
      Passed unanimously


 From sv-bc

 V-1364
Dave - these should be changed to sv-bc

 0000999  check index
	  Index for 1364 LRM

      On March 13, 2006, the SV-BC voted to resolve this issue without
      addressing the specified items. The vote was not unanimous:

      For: Francoise, Mark, Doug, Gordon, Dave, Stu
      Opposed: Steven (no issue for index in SV-BC)
	       Shalom(no issue for index in SV-BC)
	       Brad(no issue for index in SV-BC)
      Karen (leave it the way it is)

      A new svdb entry is to be created requesting an accurate, updated index
      in the next draft of the 1800 LRM, especially one merged with 1364.

      Issue 1643 has been filed requesting an index in the next draft of the 
      P1800 LRM.

    Shalom - Back then people didn't want the issue to be lost for 1800
           - The new issue was opened.

        Move: Shalom - mantis item 999 should be closed 
      Second: Dave
      Passed unanimously


 0001004  5.1.13: Zero fill in ?: even if signed or x/z
	  unanimously approved 06/25/2007

        Move: Shalom - approve the proposal for mantis 1004
      Second: Francoise
      Passed unanimously


 0001101  17.1.1: not clear how "\a" is interpreted (in $display)
	  unanimously approved by email vote  06/11/2007

       Shalom - A string argument changed to a string literal argument

        Move: Shalom - approve the proposal for mantis 1101
      Second: Dave
      Passed unanimously


 0001143  allow unsized numbers and integer variables in concatenations
	  Resolution: no change required
	  06/11/2007 
 
          Opposed: Shalom (would be a useful capability in concatenations)
                   Don (wants integer to be recognized as a 32-bit value)

    Shalom - His wasn't a strong opposition. 
	   - The next to last bugnote is not quite correct (3/29/07)
	   - Unsized constants are 32 bits.

        Move: Dave - approve the resolution for mantis 1143
      Second: Brad
     Abstain: Shalom - same reason he opposed in svbc
      Passed with one abstain

 0001153  parameterized task/function extensions
	  duplicate of 696

        Shalom - They were filed in parallel since 1800 and 1364 were separate.
	       - When the groups the were merged it became one issue.

        Move: Dave - approve the resolution for mantis 1153
      Second: Brad
      Passed unanimously


 0001198  Support a container to define how to interface to a set of signals.
	  duplicate - already covered by lrm

      Shalom - It was filed on 1364 but was already in the 1800 LRM.

        Move: Dave - approve the resolution for mantis 1198
      Second: Shalom
      Passed


 SystemVerilog

 0000227  Ambiguous phrase "packages must exist"
	  Unanimously approved 6/25/2007

      Stu - I do not like the change because it requires file order dependency 
	    in compilation.  I assume the BC has discussed this and decided it 
	    is not practical to avoid file order dependency.  The change does 
	    fix the ambiguous wording in the current LRM, even if I don't like 
	    the fix.

        Move: Dave - approve the proposal for mantis 227
      Second: Shalom
      Passed unanimously


 0000915  Size wrong in comments of an example of 4.10
	  Unanimously approved 6/25/2007

        Move: Dave - approve the proposal for mantis 915
      Second: Shalom
      Passed unanimously

 0000918  'medal' reference refers too far back in text
	  Unanimously approved 6/25/2007

	  Shalom - an example versus the text were different

        Move: Francoise - approve the proposal for mantis 918
      Second: Shalom
      Passed unanimously

 0000965  Name from example should use constant-width typeface (6.3.2.1)
          Unanimously approved by email vote 06/11/2007
	  Resolution: no change required

        Move: Shalom - approve the resolution for mantis 965
      Second: Brad
      Passed unanimously


 0001064  Multi-line string literals?
	  Unanimously approved 6/25/2007

       Brad - affects backward compatibility?
     Shalom - yes
	    - Doesn't change what is in the standard
	      Some simulators behave differently
	      Vendors agreed to match what was in the standard
	    - Checks of regression databases didn't find any user examples
	    - Currently defined LRM is preferable even if tools need to be 
	      changed. 
	    - Some tools were not following the text in 1800-2005 

        Move: Dave - approve the proposal for 1064
      Second: Francoise
      Passed unanimously

 0001297  genvar clarification
          Proposal approved unanimously in the 9/21/06 P1800 WG Meeting.

          Stu wasn't sure what changes were required [Mantis "Additional 
	    information field]

	  During the March 13, 2006 SV-BC meeting this issue was unanimously
	  voted as resolved because it was deemed addressed in 1364.

     ---> This resolution means that no further action is required by the 
          editor.

        Move: Shalom - move that we approve the resolution for 1297
      Second: Brad
      Passed unanimously


4. Meetings 

   August 15
   August 29 9am pst



List of action items from this meeting

1. Neil - send mantis 1704 back to the SV-AC for the changes mentioned in 
          the meeting.
2. Shalom - add another bugnote to 1143 to clarify the existing bugnotes.
3. Neil - send email to the committees on the schedule and recommend  not holding
	  off until the last minute for passing proposals. Otherwise the 
	  Champions will have a tough time completing their review in time.
Received on Sat, 18 Aug 2007 19:28:21 -0700

This archive was generated by hypermail 2.1.8 : Sat Aug 18 2007 - 19:28:43 PDT