Re: Attributes allocated, associated, & data_location wrt TAG interface

[Michael Eager <eager at eagercon dot com>]

Ron 603-884-2088 wrote:
> 
> In the course of editing the text for item 000323.2 re Java interface support
> into the working document, I noticed that we did not consider yesterday
> whether the attributes DW_AT_allocated, DW_AT_associated and DW_AT_data_location
> should apply to DW_TAG_interface_type (specifically in Appendix 1).
> 
> You will recall that these attributes were added to DWARF as part of item
> 991108.11 which focused mostly on array issues. However, all three attributes
> were deliberately defined to apply to all types. Of course, at the time
> of that adoption the DW_TAG_interface_type did not yet exist.
> 
> I plan to include DW_AT_allocated, DW_AT_associated and DW_AT_data_location
> in the list of attributes applicable to DW_AT_interface on the grounds that
> that is most consistent with the intent of 991108.11.

I think that the intent of 991108.11 was to represent Fortran allocated
arrays, not some broader purpose.   Any attribute can be used with any tag, but
not all of these combinations are meaningful.

As Dave W. pointed out, the list in the Appendix 1 is just advisory, neither
prescriptive or proscriptive.  They are what we would expect to see as attributes
for a given tag.  Since DW_TAG_interface_type was created to describe Java
interfaces, and since the three attributes added by 991108.11 (as well as many other
attributes) have no clear meaning in this context, we wouldn't expect to see
them.  That isn't to say that someone at some future date with some future
language wouldn't use them, just that they have no meaning with the currently
supported languages.  

The list that we came up with at the meeting on Tuesday seems to be the 
list of attributes which we would expect to find with DW_TAG_interface_type.

-- 
Michael Eager	 Eager Consulting     eager@eagercon.com
1960 Park Blvd., Palo Alto, CA 94306  650-325-8077