P1076 December 1, 2016 Meeting Minutes

Attendees:

• Brent Hayhoe, Peter Flake, Rob Gaddi, Jing Pang, Jim Lewis, Patrick Lehmann, Kevin Jennings, Lieven Lemiengre

Agenda:

Meeting Discussion

• LCS Review
• By December, need LCS for proposals that will go into standard.
• December to March - prepare final document.
• Format of LCS
  • Common format
    • Red Strike Through for deleted text
    • Black for unchanged LRM Text
  • format 1:
    • Blue underline for new text
  • Format 2:
    • Red for new text
  • Resolution:
    • allow 1 and 2 for existing LCS
  • Going forward:
    • Format 1:
    • Format 2: Kevin, Rob, Peter, Jim
    • Abstain: Brent (happy either)
• Shared Variables of ordinary types
  • Synthesis & dual port RAM with asynchronous clocks
    • Historic synthesis tool support did this
    • Current tools support signals in a single process with multiple clocks.
  • Ease of initialization - but there are plans to fix this in 2017
  • Not necessary: Patrick, Rob, Brent, Lieven, Jim
• ISAC must do?
  • Inconsistent and/or incorrect things in LRM
• AI: Jim contact John Shields and Steven Dovich about P1735 updates
  • Can we get more by reference and less in LRM?
• Extended Enums
  • There is nothing like this in the current requirements list
  • Patrick: ADA has something like this? Only tagged records and not enumerations.
  • Lieven: Typical enumerations are a closed set, what is being asked for is an open set.
  • Would require someone to work on it.
• Type Equivalence
  • From the email:
    • A means of stating that two separate type definitions are the same (MAYBE even if they are not exactly the same)
      a) Simple case: E.g. allow a byte_type to be connected to a byte_type even if these are defined in two separate packages.
  • Analysis:
    • Note: if both types are visible, they can currently be type casted.
    • is byte_type a type or a subtype?
    • if each byte_type were given a unique name (ie: byte_type1 and byte_type2) and they were made subtypes of the same type, such as std_logic_vector, then they will already automatically convert.
  • With type generics, this may happen more often.
  • Implicit conversions between two types that are structurally the same:
    • IE: 2 records
  • Sounds similar to closely related records,
    • however, in that proposal position, name, and type of the elements must be the same
    • could closely related records be extended to also handle closely related arrays.
  • May be able to handle this with a space ship operation if done element by element.
  • May be supported by an already existing proposal, however, if not, there is no bandwidth to consider it at this time.

Review and Approve Meeting Minutes and Decisions by Attendees

Lieven, Rob, Peter, Kevin, All

Review and Approve Meeting Minutes and Decisions by non-attendees

TBD

Next Meeting: Thursday December 8, 2016, 11 am Pacific

Previous Meeting: Thursday November 24, 2016