[sv-bc] Mandated warnings -- was Re: [sv-ec] Mantis 2701, ballot ID #44 - Arturo's feedback

From: Gordon Vreugdenhil <gordonv_at_.....>
Date: Fri May 01 2009 - 07:01:16 PDT
jonathan.bromley@doulos.com wrote:
[...]

>> As to the warnings, this proposal
>> is too restrictive to vendors, and the LRM generally gives 
>> wider latitude to implementations with respect to warnings.
> 
> Yes, other people have said the same.  As long as everyone is
> happy to mandate at least one warning, then it should be easy
> to relax and simplify that part of the proposal.


I wouldn't be happy about it, but I probably wouldn't oppose
the change for that reason as the weaker statement likely
still gives sufficient flexibility.


I am very concerned about the trajectory of the standard in
terms of "warnings".  The recent vote in SV-BC where vendors
unanimously opposed a warning on any array bound write violation
and users unanimously supported the warning is a good indicator.

Vendors do not go out of their way to make life miserable
for users.  We know that warnings about surprising results
can be useful; in fact, vendors in general have a vested
interest in having reasonable and predictable warnings since
it can help to reduce support interactions.  At the same
time, we *regularly* see users turn off *ALL* warnings since
they just don't want to see the noise.  Good or bad, that
is reality.

So when *mandated* warnings are suggested that might impact a
vendor's ability to do aggressive optimizations, I will
object.

I would ask the users to keep that in mind -- we know that
users want consistency, but consistency can bear a pretty
significant *real* cost in terms of performance and/or
a vendor's ability to innovate and be aggressive.

If the response is "turn the warnings off to be aggressive",
then I don't think the warning should be mandated.  Mandating
behavior that is *expected* to be violated in most real
situations is not good language definition practice and I won't
support it.

Gord
-- 
--------------------------------------------------------------------
Gordon Vreugdenhil                                503-685-0808
Model Technology (Mentor Graphics)                gordonv@model.com


-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.
Received on Fri May 1 07:01:59 2009

This archive was generated by hypermail 2.1.8 : Fri May 01 2009 - 07:02:17 PDT