ALF meeting Feb. 2, 2000

Pierre Girouard		Logicvision
Greg Dufourd		Mentor Graphics
Shir-Shen Chang 	Synopsys
Priti Vijayvargiya	Synopsys
John Peters		Philips
Tim Jennings		Philips
Praveen Kashyap		Motorola
Simon Favre		Monterey
Anand Sethuram  	LSI Logic
Antonio Lagnada 	LSI Logic
Dan Moritz      	LSI Logic
Ekabaram Balaji 	LSI Logic
Naveen Gupta    	LSI Logic
Dhaval Sejpal   	IBM
John Beatty     	IBM
Paul Zuchowski  	IBM
Wolfgang Roethig	NEC
Jay Abraham     	SI2

In the afternoon, the following people joined:
Kent Moffat     	Mentor Graphics
Cary Wei        	Fujitsu

1) Review of ALF spec. 2.0

See ALF spec on the web and Mike Andrew's email with comments.

* CELLTYPE
Need statement of primary purpose
Suggestion: The purpose of CELLTYPE is to classify
the function of cells in broad categories.

Celltype PAD does not fit into this description and
should be rather defined as PLACEMENT_TYPE for
layout modeling.

Celltype "block" and "core" classify complex cells.
A CELL with CELLTYPE "block" should have a STRUCTURE
statement, whereas "core" is a monolythic hardmacro.

Issue for DFT: How are boundary scan cells classified?
A.I. Describe functionality of boundary scan cells - Pierre Girouard

A.I. propose more celltypes for complex cells
     such as PLL, CPU etc. if required - all

* multiple non-scan-cells
multi-value-assignment is the prefered syntax.

* POLARITY for output signal
to be removed from spec., because overloaded,
not required, inconherent.

* enable_pin
Only case where an annotation inside a PIN refers to another pin.
Annotations inside a pin should be within the scope of the pin itself.
Relationships between pins should be described by other statements.
It is recommended to use the new OPERATION statement to describe
the output-enable operation.

* PRIMITIVE definition in FUNCTION
to be removed from spec., PRIMITIVE definition in CELL is sufficient.

* Amended semantics for SIGNALTYPE
DATA	o.k.
TIE	o.k.
CONTROL	o.k.
ADDRESS	o.k.
CLOCK	Proposed changes: change "data" to "information"
	Add statement "A clock is a synchronizing signal for sequential element".
ENABLE	o.k., be more precise about POLARITY in enable/disable case.
SET	o.k.
CLEAR	o.k.

* PURPOSE annotation or CLASS
"iddq" is OPERATION or PURPOSE?

* SUPPLY_CLASS
o.k.

* PHYSICAL_MAP
This keyword is needed, the mapping inforamtion does not belong in FUNCTION.
A.I. explain the physical address description mechanism - Naveen Gupta and Pierr Girouarde

* Address out of range
MIN/MAX construct is not sufficient, because the address space can have
"holes" in the middle.
Use ILLEGAL as keyword instead of CONFLICT.

A.I. set up DFT conference call next week - Mike Andrews

* KEYWORD
o.k., but need to disallow illegal redeclaration of already standardized keywords

A.I. update ALF spec 2.0 - Wolfgang Roethig

2) Discussion on OLAWorx usage and OLA support - see OLA minutes