Re: [Boost-build] rtti patch

From: Jaroslav Gresula (jgresula.LEAVE-THIS-OUT_at_[hidden])
Date: 2008-02-16 11:10:36

Hi Jurko,

Jurko Gospodnetić wrote:
>    Hi Jaroslav.
>
>> it looks like '<rtti>off' does not turn off rtti for gcc and msvc-8.0
>> toolsets.
>> 
>> gcc toolset does not handle rtti feature at all which causes that rtti
>> is generated by default. msvc toolset handles only <rtti>on, but rtti
>> is turned on by default (at least since 8.0).
>
>    Not a gcc expert and did not take the time to play around with this, 
> but should not the -frtti flag be passed to gcc explicitly? Just so we 
> do not depend on the default compiler behaviour?
>
>    I did do 'gcc -v --help' on my cygwin gcc installation and it listed 
> the -frtti option but no -fno-rtti. Many other -fno-XXX options are 
> listed expliclitly so I assume there is no general rule that each -fXXX 
> option has its corresponding -fno-XXX option.

I'm not a gcc expert either. I have two gcc versions available, 4.1.2
on Linux and 3.4.4 on Cygwin and both accept -fno-rtii. It seems that
this option is supported since early gcc versions according to

  http://gcc.gnu.org/onlinedocs/gcc-2.95.3/gcc_2.html#SEC3

>    Similar for msvc - should we change it to set /GR or /GR- based on 
> the current <rtti> setting and not depend on the default compiler 
> behaviour at all? This semms like it would work the same on all 
> supported compiler versions and there would be no need to version 
> specific code.

Good idea.

-- 
Jarda