[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

--- Comment #6 from Richard Biener  ---
Note I didn't followup on my proposal as I didn't have time to compare
bootstrap times (on a non-multilib platform it will likely even regress because
in stage3
we build both the host and the target libstdc++).  The proposal also has gotten
a "hole" since we now bootstrap libsantizer (another waste of ressources)
because
of bootstrap-{a,ub}san and friends.  That one as well should be a non-bootstrap
target module and bootstrap-{a,ub}san should add a host module for
libsanitizer.
I'm also sure that libvtv doesn't need bootstrapping if libstdc++ becomes a
target module.

That is, we have to watch dependences between these modules.

Hmm, do we even build host modules in stage3? ...  I think we don't need them.

[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[redi at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

--- Comment #5 from Jonathan Wakely  ---
No objections from me. I'm not sure how representative of real C++ code the
bits in libstdc++.so are anyway.

[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[egallager at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

Eric Gallager  changed:

   What|Removed |Added

 Status|UNCONFIRMED |NEW
   Last reconfirmed||2017-07-22
 CC||egallager at gcc dot gnu.org
 Ever confirmed|0   |1

--- Comment #4 from Eric Gallager  ---
(In reply to Richard Biener from comment #2)
> (In reply to comment #1)
> > Well, for profiledbootstrap I think it is nice if target-libstdc++-v3 is
> > actually bootstrapped, because then we are able to train the C++ FE on real 
> > C++
> > code.
> 
> Well, we bootstrap the host-libstdc++, that should be enough, no?
> 
> Bootstrapping libstdc++ multilib and with building the PCHs is excessive
> waste of resources for building GCC with a C++ compiler.

Confirming that bootstrapping libstdc++ with multilibs is an excessive waste of
resources.

[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

Jakub Jelinek jakub at gcc dot gnu.org changed:

   What|Removed |Added

 CC||jakub at gcc dot gnu.org

--- Comment #1 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-06 
12:58:35 UTC ---
Well, for profiledbootstrap I think it is nice if target-libstdc++-v3 is
actually bootstrapped, because then we are able to train the C++ FE on real C++
code.

[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

--- Comment #2 from Richard Guenther rguenth at gcc dot gnu.org 2012-03-06 
13:15:37 UTC ---
(In reply to comment #1)
 Well, for profiledbootstrap I think it is nice if target-libstdc++-v3 is
 actually bootstrapped, because then we are able to train the C++ FE on real 
 C++
 code.

Well, we bootstrap the host-libstdc++, that should be enough, no?

Bootstrapping libstdc++ multilib and with building the PCHs is excessive
waste of resources for building GCC with a C++ compiler.

[Bug bootstrap/52509] target libstdc++-v3 should not be bootstrapped, libstdc++-v3 should also be a host_module (bootstrapped)

[jakub at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=52509

--- Comment #3 from Jakub Jelinek jakub at gcc dot gnu.org 2012-03-06 
13:31:35 UTC ---
I guess bootstrapping of the host libstdc++-v3 if it is performed is fine.