RE: [sv-cc] vpiName

The quick, jocular answer is because some things have names but
don't have full names.
 
Historically, an object was supposed to have a full name if
you could use that full name to find the object using
vpi_handle_by_name(<full_name>, NULL).  With the advent of
SystemVerilog, we now have objects that don't have a fixed
relationship to the design root, or perhaps a relationship
that we've not yet decided to codify.
 
Let's consider the in-flight vpiClassTypespec objects, for example.
The proposal says that if the vpiClassTypespec represents a class
specialization, it shall a valid though tool-dependent name.
If that tool-dependent name is the same as the name of the
corresponding vpiClassDefn, then we have an ambiguity if
we try to embed that name in a full name.  I think we probably
could, but at least one of the vendors (namely, the one I
work for) doesn't yet have enough experience to lock down
on a particular choice.  And even if we did, we'd have to
convince the other vendors.  That probably won't happen in
the current round.
 
For a vpiTypespecMember, the vpiName is the identifier
for that member, but it does not (in this context) name
a particular member of a particular struct, and so does
not have a full name.
 
Similarly, we have not yet figured out how to specify
a full name for typespecs in general.  We probably could,
but then what would one use it for?
 
Again, we don't within the normal VPI use model provide
a vpiFullName for a task or function call.  Part of the
reason for this is that VPI allows you to go from
a context (such as a containing statement or other
expression) to an expression, but we're not set up
to compute the reverse relation.  And if multiple
instances of a function call are embedded in the
same containing other expression, what kind of
full name could you construct?  But it does make
sense to return the identifier of the corresponding
task or function declaration.
 
So I expect it's just a buncha different reasons.
 
Regards,
Jim Vellenga

---------------------------------------------------------
James H. Vellenga                            978-262-6381
Software Architect                              (FAX) 978-262-6636
Cadence Design Systems, Inc.         vellenga@cadence.com
270 Billerica Rd
Chelmsford, MA 01824-4179
"We all work with partial information."
---------------------------------------------------------- 

 


________________________________

	From: owner-sv-cc@eda.org [mailto:owner-sv-cc@eda.org] On Behalf
Of Bresticker, Shalom
	Sent: Tuesday, August 26, 2008 9:45 AM
	To: SV-CC
	Subject: [sv-cc] vpiName
	
	

	Hi, 

	In the VPI diagrams, why does vpiName sometimes appear without
vpiFullName as well? 

	Thanks, 
	Shalom 

	Shalom Bresticker 
	Intel Jerusalem LAD DA 
	+972 2 589-6582 
	+972 54 721-1033 

	
---------------------------------------------------------------------
	Intel Israel (74) Limited
	
	This e-mail and any attachments may contain confidential
material for
	the sole use of the intended recipient(s). Any review or
distribution
	by others is strictly prohibited. If you are not the intended
	recipient, please contact the sender and delete all copies.

	-- 
	This message has been scanned for viruses and 
	dangerous content by MailScanner <http://www.mailscanner.info/>
, and is 
	believed to be clean. 


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