Re: [boost] Boost.Build/Boost.Jam issues w/ gcc v302

From: David Abrahams (dave_at_[hidden])
Date: 2002-08-14 14:44:17

From: "Ronald Garcia" <garcia_at_[hidden]>

> Ronald Garcia wrote:
>
> > My R of TFM  (which happened before I asked :) ) seems to suggest that
> > "-I" will  suffice (info pages for the gnu C preprocessor). Is there a
> > difference? (besides breaking on my architecture ;) )
>
> In answer to my own question, it turns out there *is* a difference, but
> it's sure funky.
>
> Using gcc to preprocess code under solaris shows that one thing changes
> in the output:  The line directives.
> Line number information includes a line number a filename and flags.
> One flag indicates that "the following text comes from a system header
> file, so certain warnings should be suppressed." This is indicated by
> the flag number "3"
>
> Another says "the following text should be treated as C."  This is flag
> number "4"
>
> using -isystem is adding both the 3 and 4 flags to line number
> information.  Now, perhaps the config library needs the "3" flag for
> something under linux, but for some reason both the "3" and "4" added to
> the output under solaris cause problems.

Well, it's a GCC (or preprocessor) bug.
Have you reported it through GNATS?

-----------------------------------------------------------
           David Abrahams * Boost Consulting
dave_at_[hidden] * http://www.boost-consulting.com