Mailing List picongpu-users@hzdr.de Message #146
From: René Widera <r.widera@hzdr.de>
Subject: Re: [PIConGPU-Users] configure script acts different when restart
Date: Fri, 11 Nov 2016 09:46:36 +0100
To: <picongpu-users@hzdr.de>
Dear Danila,

the C++11 compile issue is now solved and merged to the release candidate 0.2.0 [1].
Ax3l will release PIConGPU 0.2.0 today.

best reguards,

René Widera

[1] https://github.com/ComputationalRadiationPhysics/picongpu/tree/release-0.2.0

On 10/12/2016 03:34 PM, Khikhlukha Danila wrote:
Dear all,

It looks like I have found a solution although it is not very
straightforward and looks more like a workaround. I would still ask you
to comment on it a bit. Please refer to the script attached to this
email. With this script I was able to compile picongpu binary.

1.       To pick the correct version of g++ I had to show cmake the path
to compiler explicitly (export CXX and export CC from the script
attached.) Not sure why cmake didn’t pick it up from the PATH, although
`which g++` works fine.

2.       Somehow the env var |export CPPFLAGS="-std=c++11" |from Install
manual (mandatory section) doesn’t work.Better to say it changes nothing
as I still get an error from the boost library. However when I changed
it to “export CXXFLAGS” (line 11) the error about __float128 is gone,
 which is expected.

3.       If I do configure && make && make install I receive a message
“nvcc fatal   : redefinition of argument 'std'”. It seems like during
“make install” stage the binary is being recompiled and std flag is
being include to the complier’s command line once again. However
configure && make install (line 16) works just fine.

With the script attached I was able to compile the picongpu binary from
the 0.2.0 branch.



Best regards,

Danila.



*From:* Khikhlukha Danila
*Sent:* Tuesday, October 11, 2016 11:50 AM
*To:* picongpu-users@hzdr.de
*Subject:* configure script acts different when restart



Dear all,

I’m not sure if it is a bug or I’m just doing it wrong, so I decided to
write here. However I can move this letter to the bug tracker.



Configure script acts different when restart in the same build
directory. I would consider it as a wrong use model unless it wouldn’t
actually do the job J



Scenario 1.

Trying to build picongpu from the scratch (makeit script and
log_makeit). Compilation fails due to the boost error with __float128.
Please notice that although I have explicitly loaded gcc 4.9 version,
cmake still thinks that gcc 4.8.5 is used (loading this module from
bashrc doesn’t change it).



Scenario 2.

I “forget” to provide C++11 flag  (makeit_restart and
log_makeit_restart). First compilation fails asking for a C+11 flag.
Then I immediately correct my mistake and restart configuration and
compilation in the same directory.  This works like a charm. From the
log it is evident that somehow after restart a different compiler
version was picked. (also note that the build type was changed to a
DEBUG version for some reason.)



Scenario 3.

If I clean the build directory before the restart (makeit_restart_remove
and log_makeit_restart_remove) it will be the same as Sc.1,  same
compiler version and the very same boost error.



I checked the CMakeCache.txt between Sc1. and Sc2. before the restart
and didn’t find any relevant difference. Could you please advise me how
can I fix the Sc.1?



Best regards,

Danila.



P.S. Please note that I don’t have an access to the boost config file
(https://svn.boost.org/trac/boost/ticket/11852)



#############################################################
This message is sent to you because you are subscribed to
  the mailing list <picongpu-users@hzdr.de>.
To unsubscribe, E-mail to: <picongpu-users-off@hzdr.de>
To switch to the DIGEST mode, E-mail to <picongpu-users-digest@hzdr.de>
To switch to the INDEX mode, E-mail to <picongpu-users-index@hzdr.de>
Send administrative queries to  <picongpu-users-request@hzdr.de>


--
René Widera
Abteilung Laser-Teilchenbeschleunigung (FWKT)
Helmholtz-Zentrum Dresden-Rossendorf
Tel: +49 (0351) 260 3543
r.widera@hzdr.de
http://www.hzdr.de

Vorstand: Prof. Dr. Dr. h. c. Roland Sauerbrey,
          Prof. Dr. Dr. h. c. Peter Joehnk
Vereinsregister: VR 1693 beim Amtsgericht Dresden
Subscribe (FEED) Subscribe (DIGEST) Subscribe (INDEX) Unsubscribe Mail to Listmaster