[sv-cc] Testing changes for 1865

One of my action items was to test the effects of the
changes proposed for Mantis item 1865.  I have now
done that, running the changes against our most
commonly used regression suite using the EE3.0
operating system.

Most of the failures were due to the changes in
DPI (which are not yet part of the standard) in
names of the fields from "a" and "b" to "aval"
and "bval".  It was straightforward if tedious
to make these changes, as the compiler always
highlighted the file and line number.

(These changes, of course, were already required
by a previously approved proposal.)

We did have one VPI problem where an assignment
was assuming that aval was a signed integer --
something like the following:

  realVal = (double)vecVal.aval;

This now produced a wrong result without any
warning other than the failure of the regression.
(Scary.)  Once I found the problem, however,
I easily fixed it by adding yet another coercion:

  realVal = (double)(int32_t)vecVal.aval;

Summary of my findings:

DPI changes -- many, but easy to find, easy to fix.

VPI changes -- few, dangerous and harder to find, easy
  to fix.

Charles, at an upcoming meeting we should note
that this action item has now been completed.

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." 
---------------------------------------------------------- 

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