To build mfront on FreeBSD-10.2, with clang 3.4.1, the attached patches are needed.
I have'nt added any "#ifdef FreeBSD" or things like that, because that should'nt be harmful for the other platforms.
Thank you for your patches and your interest to MFront.
The problem seems due to the libc++ library which seems to automatically enable C++11. With libstdc++, no problem. The joy of portability....
However, tfel shall compile even if C++11 is enabled (which is normally the case with gcc, by this is surely a good fortune : maybe libstdc++ headers that are used do not include the <memory> header internally...), so this is a bug.
The "good news" are that I was able to reproduce it on my Debian box. Your corrections have been integrated to the 2.0.x and 2.1.x branches which will be available from sourceforge soon (our internal jenkins server must first compile and passe all testes before exporting the sources to the sourceforge servers, please check the Devel/ section in a few hours). I would appreciate if you could test it.
Hopefully, we don't have the same issues with the 3.0.x branche which is fully based on C++-11 (tfel::utilities::shared_ptr have been removed) and has been recently ported to PC-BSD and Mac OSX which both use clang/libc++ by default. Indeed, I was surprised that you did not have some other issues with libc++ as it is very touchy about the std::complex class, but those troubles have also been handled in all the development branches.
FreeBSD being an usual platform for us (though we wish to fully support it), please consider reporting your feed-back.
Sincerly,
Helfer Thomas
tfel-contact@cea.fr
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
Le mar 25 aoû 15 à 11:07:45 +0200, Helfer Thomas thelfer@users.sf.net
écrivait :
Dear Thierry,
Hello!
Thank you for your patches and your interest to MFront.
Actually, I did not know anything about MFront, but I am the maintainer
of the Code_Aster's port for FreeBSD, and it is bundled within the
source package. It seems a very interesting program!
The problem seems due to the libc++ library which seems to
automatically enable C++11. With libstdc++, no problem. The joy of
portability....
However, tfel shall compile even if C++11 is enabled (which is
normally the case with gcc, by this is surely a good fortune : maybe
libstdc++ headers that are used do not include the <memory> header
internally...), so this is a bug.
Yes, even on FreeBSD, the problem does not allways appear, depending on
the version.
The "good news" are that I was able to reproduce it on my Debian box.
Your corrections have been integrated to the 2.0.x and 2.1.x branches
which will be available from sourceforge soon (our internal jenkins
server must first compile and passe all testes before exporting the
sources to the sourceforge servers, please check the Devel/ section in
a few hours). I would appreciate if you could test it.
Hopefully, we don't have the same issues with the 3.0.x branche which
is fully based on C++-11 (tfel::utilities::shared_ptr have been
removed) and has been recently ported to PC-BSD and Mac OSX which both
use clang/libc++ by default. Indeed, I was surprised that you did not
have some other issues with libc++ as it is very touchy about the
std::complex class, but those troubles have also been handled in all
the development branches.
FreeBSD being an usual platform for us (though we wish to fully
support it), please consider reporting your feed-back.
Great! I know that several people are using the package of Code_Aster on
FreeBSD / PC-BSD (mostly in Universities), and so they will be able to
get MFront too.
Dear Thierry,
Thank you for your patches and your interest to MFront.
The problem seems due to the libc++ library which seems to automatically enable C++11. With libstdc++, no problem. The joy of portability....
However, tfel shall compile even if C++11 is enabled (which is normally the case with gcc, by this is surely a good fortune : maybe libstdc++ headers that are used do not include the <memory> header internally...), so this is a bug.
The "good news" are that I was able to reproduce it on my Debian box. Your corrections have been integrated to the 2.0.x and 2.1.x branches which will be available from sourceforge soon (our internal jenkins server must first compile and passe all testes before exporting the sources to the sourceforge servers, please check the Devel/ section in a few hours). I would appreciate if you could test it.
Hopefully, we don't have the same issues with the 3.0.x branche which is fully based on C++-11 (tfel::utilities::shared_ptr have been removed) and has been recently ported to PC-BSD and Mac OSX which both use clang/libc++ by default. Indeed, I was surprised that you did not have some other issues with libc++ as it is very touchy about the std::complex class, but those troubles have also been handled in all the development branches.
FreeBSD being an usual platform for us (though we wish to fully support it), please consider reporting your feed-back.
Sincerly,
Helfer Thomas
tfel-contact@cea.fr
Le mar 25 aoû 15 à 11:07:45 +0200, Helfer Thomas thelfer@users.sf.net
écrivait :
Hello!
Actually, I did not know anything about MFront, but I am the maintainer
of the Code_Aster's port for FreeBSD, and it is bundled within the
source package. It seems a very interesting program!
Yes, even on FreeBSD, the problem does not allways appear, depending on
the version.
Great! I know that several people are using the package of Code_Aster on
FreeBSD / PC-BSD (mostly in Universities), and so they will be able to
get MFront too.
Truly yours,
Th. Thomas.
Hi Thierry,
do you intent to make a specific TFEL/MFront package in BSD port's system or will it be hidden in the Aster package ?
Sincerly,
Le mer 26 aoû 15 à 9:48:07 +0200, Helfer Thomas thelfer@users.sf.net
écrivait :
Hello!
At the moment, from the Code_Aster source tarball, we build several
packages: ASTK, Eficas, aster, homard, metis-edf, etc.
MFront too will be one separate package. I've not yet committed it,
because I'm still working on aster-12.4.
Regards,
Th. Thomas.
Dear Thiery,
Please let me now when you're finished. That would be a nice annoucement for the news page. Please also consider fixing Ticket #11.
Sincerly,
Helfer Thomas