Re: [Boost-users] [Proto] implementing an computer algebra system with proto

Subject: Re: [Boost-users] [Proto] implementing an computer algebra system with proto
From: Eric Niebler (eric_at_[hidden])
Date: 2009-01-28 12:03:24

Dave Jenkins wrote:
> "Eric Niebler" <eric_at_[hidden]> wrote in message 
> news:49807DB5.2080307_at_boost-consulting.com...
>> You're passing an expression to a grammar to be evaluated with its 
>> transforms. That expression must match the grammar. It's undefined 
>> behavior if it doesn't, and in this case, it doesn't. "Performing the 
>> transform" falls withing UB.
>>
>> It would be possible to add a compile-time check to all grammars 
>> function call operators to catch this error, but that comes with a 
>> compile-time cost.
> 
> I think it would be a useful check.  Could you either
> (1) always check in proto::exact

That would essentially involve a full evaluation.

> (2) check when a compile time flag is defined

That is not a half-bad suggestion. It's funny that we have NDEBUG to 
avoid costly runtime checks but no equivalent for costly compile-time 
checks.

I've never particularly cared for the "N" in "NDEBUG" ... it seems the 
wrong default. Perhaps we can have a boost-wide macro for expensive 
compile-time checks. How about BOOST_COMPILE_TIME_DEBUG?

-- 
Eric Niebler
BoostPro Computing
http://www.boostpro.com