RE: [sv-ac] P1800 SV-AC: vote on #1549

 

 

________________________________

From: owner-sv-ac@eda.org [mailto:owner-sv-ac@eda.org] On Behalf Of
Bassam Tabbara
Sent: Friday, September 08, 2006 1:54 PM
To: sv-ac@eda.org
Subject: RE: [sv-ac] P1800 SV-AC: vote on #1549 

 

I vote No. I think voting now is rushed -- we need take back to
committee for more discussions on the following:

 

a) Clarify how the new types should be handled in checking and error
reporting and also in relation to the "true nature" identification
functions that Dmitry proposed (isSequence() etc...). Currently it's not
clear at all what compiler *shall* vs. may vs. .... do if the type is
different than nature, and what the user expectation might be here ....
[Note the proposal currently has a bunch of "may be's" that bug me.]

[Lisa Piper >>>] I did a search and see 3 "may ..." and those statements
say that <things> "may" be typed.  They might also be untyped in which
case no type checking will occur.    Is this what you are referring to? 

 

How the "new types" should be handled in checking and error reporting is
a generic issue that applies to all types in all functions, tasks,
classes etc?  Do you see some conflict with Dmitry's proposal?   

b) "implicit" as we discussed before is handy potentially for SV in
general, that said, it should probably be added elsewhere (with other
types/qualifiers) and referenced here. So we should check with other
committees to find the best way to handle this (incl. double-check on an
existing/better way even) rather than add it exclusively here.

[Lisa Piper >>>] does using it here prevent it from being used
elsewhere?

 

Thx.

-Bassam.