[sv-cc] Minutes for 1/23/04 meeting


Subject: [sv-cc] Minutes for 1/23/04 meeting
From: Bassam Tabbara (bassam@novas.com)
Date: Fri Jan 23 2004 - 09:13:44 PST


Here are the minutes, I've tried to log as much as I could from the
discussion, and made sure to "bold" the action items.

Minutes: 1/23/04 (taken by Bassam):

Attendance:

- Joao

- Michael

- Francoise

- Bassam

- Ralph

* Joao propose/Bassam second approve meetings from last time

* VPI (1/21 version) Issues review:

*30.9: Variable var_select (discussion): Joao says behaves like 1364, only
applicable to array variables

- 2 different methods (on far left and far right)

- Iterate on variables (removed), one with vpiReg is for backward
compatability. Typically should use the var_select one.

- Variables are missing an iterator on index.

- Var select only usable on an array var

- Michael: User would not be concerned about 2 ways, we have 2 handles, it's
ok

- Joao:

- Joao: Idea merge the 2,so that from variables we can get to var_select.

- Francoise: The difference v[1] is a variable, v[i] is a var select, in
1364 these leads to two objects, right ?

- Joao: Not really, just the diagrams (reg and var) were split but
functionally no difference, indexing into collection.

- (action_items):

Move var select to variables

Add iterator on index: get index expressions (if any) associated with that
handle.

**Not quite backwards compatible (but this figure already breaks it: array
var already collects what 1364 had as "integer" (array of int)).

- Micheal: Must state if we do not check type of the variable array is ok.
SV you get array variable, in 1364 you get the base type.

- Francoise: List all things that are not compatible with 1364 (iteration
methods, and the types...)

- Joao to Francoise: Please send the fm file after consulting with Charles.

- Change title to variables instead of variable

* 30.34: Operators: how to represent the arguments of the delay operator.
(Add notes to explain this).

Joao: How is it handles in 1364 ? None of the operators are described...

Francoise: This is a bigger problem here... may be we can come up with a
rule ?

Joao: There is a rule (textual order), except in the cases of ranges (delay)
and repetition, always seq1, seq2 and then the modifies to the operators.

Francoise: After precedence rules are applied ...

*30.34: Expression of properties:

- back to 30.19 (action items): collapse dist item back into constrain dist
and simplify, so get an expression for value range and expression for the
weight.

* 30.18 "dist" can be merged with 30.19, get rid of dist item and then
reference back add this to 30.34.

- Add "distribution" into BNF and reference in both chapters-- send to ???

- Bassam can resend the property feedback (might get lost in the mix of
emails).

* Discussion (question from Francoise): Which "values" are we getting
(assertions sample values) and so on ? Bassam/Joao you get an object for the

semantics ...

* 30.35: (action items) Title should be "attributes" not instances.

- For every object in this diagram, make all class references to be italics
and dotted lines. (not all just some of them e.g. constraints) ??

* 30.39 Francoise: how is the slice of an array represented:

- Yes it is a var select (a simple expression).

Joao: change part select to have a property that says single or multiple
part select (with an iteration on range).

- another choice: introduce a "multiple part select".

- 3rd option: make it an operator, problem goes away.

- (action items) issue is still open.

 Joao proposes to:

** Add sliced object that has an interator on range (Francoise: correction
either range or expression).

 



This archive was generated by hypermail 2b28 : Fri Jan 23 2004 - 09:19:06 PST