Re: Attribute names


Subject: Re: Attribute names
From: Adam Krolnik (krolnik@lsil.com)
Date: Tue Nov 19 2002 - 13:22:43 PST


Hello Brad;

You ask, "why is there no way to declare attributes and their types?"

First attributes are typeless - see perl...

As to why not declare attributes, the only benefit I see is that
you would receive is that someone who mistyped an attribute would
get some error. But then, you must require all legal attributes to
be defined. Then to add more, you have to add to the definitions, etc.

You ask, "wouldn't a project be employing a limited set?" I don't know.
Right now most compilers do not support this (why, I don't know since it
has zero impact downstream, except to annotate a data structure.)

These were intended to be a simple mechanism to attach specific information
to a specific component in a design. One could devise methods where you wanted
only specific attributes and only in particular places and then have tools
to ensure compliance with this standard.

The mechanism was open ended to support a variety of tools without a lot of
restrictions. Certainly each tool could decide how they want to support their
attributes: specific names, places, objects, etc. The rest would be ignored.
That is the intent.

      Adam Krolnik
      Verification Mgr.
      LSI Logic Corp.
      Plano TX. 75074



This archive was generated by hypermail 2b28 : Tue Nov 19 2002 - 13:24:34 PST