Re: ENABLE_SYMBOLS=true does nothing

On Wed, Oct 31, 2012 at 10:35 AM, Lubos Lunak <l.lunak@suse.cz> wrote:
> On Wednesday 31 of October 2012, Norbert Thiebaud wrote:
> > On Wed, Oct 31, 2012 at 8:34 AM, Lubos Lunak <l.lunak@suse.cz> wrote:
> > > I'd prefer to not bring it back because of a special use case that's
> > > already easily doable as it is.
> >
> > Well tinderbox works nicely and easily with an autogen.lastrun
> > associated with a given profile...
> > messing with env variable is not as neat and is just a pain when
> > managing many of these profile...
> > what is the harm of --enable-symbol ?
>
>  ./configure --enable-symbol
>  ./configure CXXFLAGS='-Os -g'
>
>  Where's the difference in that?
>
> > note: I think the one that should go is dbgutil...
> > --enable-symbol => release build but with symbols
> > --enable-debug => debug build (with symbol of course)
> > extra feature (dbg-util) => --enable-debug + runtime trigger to
> > activate the 'special' stuff
>
>  How exactly do you envision a runtime trigger that activates binary
> incompatibility? Because that's exactly what makes dbgutil different from
> debug.

then make dbgutil feature not binary incompatible... I'm not c++
expert, but surely there is ways to provide these extra feature
without touching the interface.
iow having a impl_debug pointer that can be used if not null to do
extra stuff... very low cost for release and flexible for debug..
without the binary incompatibility

Norbert