VHDL Issue Number:             

Language_Version          VHDL-2002
Classification            Language Definition Problem
Summary                   Can an attribute name that denotes a function be used where a name is required?
Relevant_LRM_Sections     2.4, 4.2, 4.7, 5.1, 14.1
Related_Issues            
Key_Words_and_Phrases     attribute name, function name, resolution function
Authors_Name              Peter Ashenden
Authors_Phone_Number      +61 (8) 8339 7532
Authors_Fax_Number        +61 (8) 8339 2616
Authors_Email_Address     peter@ashenden.com.au
Authors_Affiliation       Ashenden Designs
Authors_Address1          PO Box 640
Authors_Address2          Stirling, SA 5152
Authors_Address3          Australia

Current Status:           VASG-Approved

Superseded By:

------------------------
Date Submitted:           20 January 2005
Date Analyzed:            21 January 2005
Author of Analysis:       Larry Soule
Revision Number:          3
Date Last Revised:        09 May 2005

Description of Problem
----------------------

In a number of places a name that denotes a function is required,
without the function being called. The issue is whether an attribute
name that denotes a function can be used in those contexts.

Specific cases are:

- In an attribute specification where the entity class is function.
  Here, however, the name must be a simple name, so an attribute name
  is ruled out.

- In a group declaration, where the corresponding element class in the
  group template declaration is function.  Here, however, attribute
  names are explicitly prohibited.

- In a subtype indication that includes a resolution function name.
  This is the interesting case.

2.4 requires a resolution function to be a pure function that has a
single input parameter of class constant that is a one-dimensional,
unconstrained array whose element type is that of the resolved signal.

4.2 requires that a function named in a subtype indication meet the
requirements of 2.4 for resolution functions.

14.1 describes the attribute T'value as being a function with a single
parameter of type string returning the base type of T.

If we have a resolved signal of type character, it would appear that
the attribute character'value could serve as the resolution function.
The signal could be declared as

  signal s : character'value character;

This is syntactically legal, since character'value is an attribute
name, which is an alternative for name.

So, for example, if three sources contribute the values ''', 'a' and
''' to the resolved signal, they might be assembled into the string
"'a'", passed to character'value, which would return the character
value 'a' as the resolved value for the signal.


Proposed Resolution
-------------------

I would suggest an explicit prohibition on use of an attribute name
for a resolution function name.  This would mirror the prohibition on
use of an attribute name in a group declaration.


VASG-ISAC Analysis & Rationale
------------------------------

The author's proposed resolution is accepted.

VASG-ISAC Recommendation for IEEE Std 1076-2002
-----------------------------------------------

Interpret the LRM as if the recommendation for future resisions had
been adopted.

VASG-ISAC Recommendation for Future Revisions
---------------------------------------------

Add a specific prohibition against resolution function names being
attribute names.  by changing the wording in section 2.4 from

A resolution function must be a pure function (see 2.1); moreover, it
must have a single input parameter of class constant that is a
one-dimensional, unconstrained array whose element type is that of the
resolved signal. The type of the return value...

to

A resolution function must be a pure function (see 2.1); moreover, it
must have a single input parameter of class constant that is a
one-dimensional, unconstrained array whose element type is that of the
resolved signal. The resolution function name must not be an attribute
name (see 6.6).  The type of the return value...


-------------END OF IR----------------