[sv-bc] New Mantis item 272

Matt, SV-BC,

I've posted a new (minor) erratum in the SV-BC category.
The description is as follows:

    In SV3.1a's LRM, Section 23.16.3, the following appears:

    When $writememb and $writememh write out data corresponding to
associative arrays,
    address specifiers shall be written to the output file.

    This is not complete enough. Consider an associative array indexed
by a class,
    or some other software-like object. $readmemh/b and $writememh/b
should be
    constrained such that they only accept associative arrays that are
indexed
    with integral types. In the case of an enumerated type, the address
in the
    pattern file should correspond to the enumerated type's ordinal
values.

I also uploaded a proposal to complete and clarify $readmem/$writemem
interaction with associative arrays. It is located at

    http://www.eda.org/svdb/file_download.php?file_id=240&type=bug

For convenience the proposal is duplicated here:

In 23.15.1, ADD the following blue text at the end of the second
paragraph:

            When working with associative arrays, indexes must be of
integral types.
            In case an associative array's index is of an enumerated
type, address
            entries in the pattern file are in numeric format and
correspond to the
            ordinal values of the enumerated type.

In 23.16.3, ADD the following blue text at the end of the second
paragraph:

            When $writememb and $writememh write out data corresponding
to associative
            arrays, address specifiers shall be written to the output
file. As specified in
            Section 23.15.1, associative arrays must have indexes of
integral types in order
            to be legal arguments to the $writememb and $writememh
calls.

If no one has any objections, could we get Matt to assign this item to
me, and
then submit it for an email vote?

Regards,
Doug
Received on Tue Oct 26 14:38:19 2004