RE: [sv-ec] Types, typedefs, and type parameters

From: Rich, Dave <Dave_Rich_at_.....>
Date: Wed Oct 17 2007 - 10:10:21 PDT
1500 only exists now to correct wording errors found during the
discussions that led to dropping the request for the restriction. That
has been noted in the issue.

-----Original Message-----
From: owner-sv-ec@server.eda.org [mailto:owner-sv-ec@server.eda.org] On
Behalf Of Mark Hartoog
Sent: Tuesday, October 16, 2007 8:46 PM
To: Vreugdenhil, Gordon; SV_EC List
Subject: RE: [sv-ec] Types, typedefs, and type parameters

What is your position on forward typedefs? Are forward typedefs
a type with "first-class status"? 

The reason I ask is that in Mantis 1500 there was an attempt to
limit the use of forward typedefs so that they could only be
used to declare a class handle and for no other purpose. This
proposal was never approved by the BC committee, but it does 
appear to represent a decidedly second class type status for
forward typedefs.   



> -----Original Message-----
> From: owner-sv-ec@eda.org [mailto:owner-sv-ec@eda.org] On 
> Behalf Of Gordon Vreugdenhil
> Sent: Tuesday, October 16, 2007 11:05 AM
> To: SV_EC List
> Subject: [sv-ec] Types, typedefs, and type parameters
> 
> There was some question in Monday's EC meeting as to what 
> Mentor was after in the type/typedef/type parameter space.
> 
> Mentor views all types as having the same first-class status 
> within SystemVerilog.  Essentially this means that we believe 
> that it is always valid to use a typedef or a type parameter 
> in a context in which a direct type reference is permitted.
> 
> So, for example, if you can extend a class type or construct 
> an object by using a direct class type identifier, you can 
> extend a type or construct an object by using a typedef or 
> type parameter that resolves to a class type.  The same 
> applies for all type-like contexts -- field declarations, 
> associative array index types, etc.
> 
> As with many language design principles, feature interaction 
> issues do show up as implementers apply principles to 
> functional systems.  These interaction issues need to be 
> resolved as specific issues without sacrificing language 
> principles and regularity.  Mentor agrees that there are a 
> few issues related to name resolution when adopting the 
> approach of all types as first-class; we've raised these few 
> specific issues in prior discussion and again in the name 
> resolution face-to-face.
> 
> Mentor believes that its approach to having all types treated 
> as first-class falls into the area of critical language 
> principles and as such, should not be unduly compromised.
> 
> Gord.
> --
> --------------------------------------------------------------------
> Gordon Vreugdenhil                                503-685-0808
> Model Technology (Mentor Graphics)                gordonv@model.com
> 
> 
> -- 
> This message has been scanned for viruses and
> dangerous content by MailScanner, and is
> believed to be clean.
> 
> 

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



-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Wed Oct 17 10:10:44 2007

This archive was generated by hypermail 2.1.8 : Wed Oct 17 2007 - 10:10:57 PDT