Language Change Specification for Optional Commas

Voting Results: Cast your votes here

Yes:

1. Rob Gaddi - 2016-11-17
2. Kevin Jennings - 2016-11-22
3. Morten Zilmer - 2016-12-11
4. Martin Thompson - 2016-12-12
5. Ryan Hinton - 2016-12-14 - ver 1
6. Martin Zabel - 2017-01-09 - ver 1
7. Patrick Lehmann - 2017-01-23 - ver 1
8. Lieven Lemiengre - 2017-01-27 - ver 1
9. Mark Zwolinski - 2017-02-23

No:

1. Jim Lewis - 2016-12-17 - ver 1
2. Farrell Ostler - 2017-02-12 - ver 1

3. Brent Hayhoe - 2017-02-16 Version 1 Comment below

Abstain:

1. Thomas Preusser - 2016-11-21

Details of Language Change

Changes are shown in red font.

Section 5.2.2.1 Enumeration types

Middle page 37

enumeration_type_definition ::=
    ( enumeration_literal { , enumeration_literal } [ , ] )

Section 5.3.2.1 Array types

Bottom page 44

unbounded_array_definition ::=
    array ( index_subtype_definition { , index_subtype_definition } [ , ] )
        of element_subtype_indication

Top page 45

index_constraint ::= ( discrete_range { , discrete_range } [ , ] )

Section 5.3.3 Record types

Bottom page 51

record_constraint ::=
( record_element_constraint { , record_element_constraint } [ , ] )

Section 6.5.7.1 Association Lists

Top page 81

association_list ::=
    association_element { , association_element } [ , ]

Section 6.9 Group template declarations

Middle page 93

entity_class_entry_list ::=
    entity_class_entry { , entity_class_entry } [ , ]

Section 8.4 Indexed names

Bottom page 111

indexed_name ::= prefix ( expression { , expression } [ , ] )

Section 9.3.3.1 Aggregates

Bottom page 132

aggregate ::=
    ( element_association { , element_association } [ , ] )

-- RobGaddi - 2016-16-03

Comments

The proposal is about having optional semicolons. Why is the LCS written to define optional commas?

-- Kevin Jennings - 2016-11-03

Agreed. Split the LCS into LCS2016_071a for the semicolon and TopLCS2016_071b for the commas so they can be voted independently.

-- Rob Gaddi - 2016-11-16

This would have been great in the beginning of VHDL. I am afraid, now it will just be another source of incompatibility.

-- Thomas Preusser - 2016-11-21

Patrick just pointed out to me that the optional commas make conditional compilation substantially more useful. Otherwise, if you have to conditionally compile out part of a association list you may have to jump through horrific hoops to make the commas work correctly.

Thomas, I'm not sure what you mean about incompatibility. VHDL2017 code requires VHDL2017 support to run; if your tools only support older language revisions then they won't support newer language features.

-- Rob Gaddi - 2017-01-10

@PL & @RG This is something that is going to make coding 30 seconds faster at the cost of making review significantly more difficult.

The conditional compilation argument is intriguing, but I suspect if one properly structures their conditional compilation, then there will be no compelling reason to allow this. Can someone provide an example of the claim about optional commas making conditional compilation more useful - a good place to add it is to the proposal.

-- Jim Lewis - 2017-01-28

@RG I think Thomas' point is that since tools do not support this now, and it is not a feature everyone is going to be asking for, how long do you think it will take for the feature to be relevant in simulation? Synthesis?

-- Jim Lewis - 2017-01-28

@JL Why does it make reviews more difficult?

Allowing optional trailing comas looks like this and does not decrease readability.

inst : entity work.foo
  port map (
    X => A,
    Y => B,
  );

Instance with conditional compilation and two optional ports. The conditional compilation must take care of comma insertion.

inst : entity work.foo
  port map (
`if port1 /= "" then
    X => A
`end if
`if port1 /= "" and port2 /= "" then
    ,
`end if
`if port2 /= "" then
    Y => B
`end if
  );

The same example with trailing commas allowed:

inst : entity work.foo
  port map (
`if port1 /= "" then
    X => A,
`end if
`if port2 /= "" then
    Y => B
`end if
  );

JL: [...] optional commas making conditional compilation more useful.

PL: No it's the other ways around. If we get conditional compilation, we'll need optional commas and semicolons, unless we want to blow up our code with too much conditional code directives to handle corner cases like the last item in a list has no comma or semicolon.

-- Patrick Lehmann - 2017-01-28

I agree with the comments about conditional compilation.

-- Mark Zwolinski - 2017-02-23