From: David Abrahams (dave_at_[hidden])
Date: 2003-03-07 10:30:01
Vladimir Prus <ghost_at_[hidden]> writes:
> David Abrahams wrote:
>> Vladimir Prus <ghost_at_[hidden]> writes:
>> > The first target will be build with default value <rtti>on. What about
>> > the second:
>> > 1. Should we request it with <rtti>on and get an error, or
>> > 2. Should we request if without any <rtti> setting and allow it to be
>> > built with <rtti>off
>> > I think (1) is way safer. But interested in other opinions.
>> I agree with you that it's safer. I was going to say that the only
>> other potentially viable option is:
>> 3. The <rtti>off is propagated back towards the first target by the
>> second, much like a use-requirement.
>> However, as I consider it, I think there's a real use-case for "I
>> know what I'm doing; I want <rtti>on on the first target, and I just
>> won't do anything evil with the library which causes me to need
>> <rtti>on with it."
> I'm sorry, I don't understand you. What is "won't do anything evil"
What's evil clearly depends on the compiler. One likely example would
be attempting to use typeid(X) on a class X defined in the library
built with <rtti>off.
> and what's "need <rtti>on with it"?
In other words, if I *did* use typeid(X), I would need lib to be
built with <rtti>on.
-- Dave Abrahams Boost Consulting www.boost-consulting.com
Boost-Build list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk