[GenABEL-dev] jenkins gcc to g++

[L.C. Karssen]

Hi Alvaro,


On 18-02-14 17:56, Frank, Alvaro Jesus wrote:
> Hi Lennart, 
> 
> thank you for the quick response, it is interesting to note that using
> 
> g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -Wall -pedantic -Wunused-result -Wmaybe-uninitialized
> 
> wont yield the warnings from the "-Wmaybe-uninitialized" flag, but adding the optimizations "-02" will:
> 
> g++ *.cpp -L/usr/lib/openblas-base -lpthread -lopenblas -llapacke -fopenmp -O2 -Wall -pedantic -Wunused-result -Wmaybe-uninitialized
> 
> O2 seems to trigger the warnings only when present.

That's interesting! Learned something new. I guess that when using -O2
the optimisations may be such that it cannot guarantee certain
initialisations.
You probably know more about compiler optimisations, but my
understanding is that -O2 is sort of the least -O flag you should use if
you want to make use of the capabilities of current hardware.
Maybe it's worth trying to profile your code with and without the -O2
option to see what the difference in speed is.

By the way, the -O2 is added in configure.ac (if the user didn't specify
CPPFLAGS when running ./configure.


Best,

Lennart.

> 
> -Alvaro
> 
> 
> ________________________________________
> From: genabel-devel-bounces at lists.r-forge.r-project.org [genabel-devel-bounces at lists.r-forge.r-project.org] on behalf of L.C. Karssen [lennart at karssen.org]
> Sent: Tuesday, February 18, 2014 5:49 PM
> To: genabel-devel at lists.r-forge.r-project.org
> Subject: Re: [GenABEL-dev] jenkins gcc to g++
> 
> Hi Alvaro,
> 
> Thank you for looking into this!
> 
> On 18-02-14 17:16, Frank, Alvaro Jesus wrote:
>> Hi all,
>>
>> I have been trying to replicate Jenkins behavior locally to avoid
>> unnecessary numbers commit>bugfix that could be committed at once. but
>> the project locally ahs to be compiled using g++ and not gcc. Jenkins
>> uses gcc for the unused variables warnings.
> 
> If I understand you correctly, you mean that the output of the "GNU Make
> + GNU C compiler warnings" is run by gcc instead of g++. Right?
> I didn't check in detail which compiler is used by this check, but in
> the Jenkins config it says that it uses the console log (see URL in the
> next sentence).
> 
> If I look at the console output of the last build
> (http://www.karssen.org/jenkins/job/OmicABELnoMM/18/console) I see the
> same warnings pop up. For example:
> 
> g++ -DHAVE_CONFIG_H -I. -I./src   -Wall -fopenmp -g -O2 -MT
> src/Algorithm.o -MD -MP -MF $depbase.Tpo -c -o src/Algorithm.o
> src/Algorithm.cpp &&\
>         mv -f $depbase.Tpo $depbase.Po
> src/AIOwrapper.cpp: In static member function ‘static void*
> AIOwrapper::async_io(void*)’:
> src/AIOwrapper.cpp:227:82: warning: ignoring return value of ‘size_t
> fread(void*, size_t, size_t, FILE*)’, declared with attribute
> warn_unused_result [-Wunused-result]
>                  fread (tobeFilled->buff, sizeof(type_precision),
> size_buff, fp_Y);
> 
> So I'm not sure if there is a problem with the "Gnu Make + ..." check.
> 
> What the Jenkins build does is the following (Jenkins runs on an Ubuntu
> 13.10 machine, so look there for the exact package versions of automake,
> gcc, etc.):
> 
> - Check SVN every 15 minutes for changes
> 
> - If there are changes, run the following:
> 
> - Remove files not in SVN (e.g. Makefile.in and .o files from previous
> builds)
> 
> - Count lines of code (sloccount) and run cppcheck:
> sloccount --duplicates --wide --details src/ | grep -vP
> ".svn|cppcheck-result.xml|sloccount.sc" > sloccount.sc
> cppcheck --quiet --enable=all --xml src/ 2> cppcheck-result.xml
> 
> - Run cpplint.py:
> #!/bin/bash -x
> cpplint.py
> --filter=-whitespace/tab,-whitespace/comments,-whitespace/newline,-whitespace/braces,-build/header_guard,-build/include
> src/*.{cpp,h} 2> src/cpplint.log
> exit 0
> 
> - Do the actual build:
> autoreconf -i
> if [ -f Makefile ]; then
>    make distclean
> fi
> ./configure --disable-silent-rules LDFLAGS="-L/usr/lib/openblas-base"
> make -j2
> 
> - Scan for compiler warnings ("Gnu Make + ...").
> - Publish cppcheck results
> - Publish sloccount results
> - Report violations from cpplint.log
> 
> 
> Best,
> 
> Lennart.
> 
>> Is there a way to perhaps change
>> https://jenkins.genabel.org/jenkins/job/OmicABELnoMM/ to check using g++
>> for more consistent results?
>>
>> Thank you for the help.
>>
>> -Alvaro
>>
>>
>> _______________________________________________
>> genabel-devel mailing list
>> genabel-devel at lists.r-forge.r-project.org
>> https://lists.r-forge.r-project.org/cgi-bin/mailman/listinfo/genabel-devel
>>
> 
> --
> *-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
> L.C. Karssen
> Utrecht
> The Netherlands
> 
> lennart at karssen.org
> http://blog.karssen.org
> GPG key ID: A88F554A
> -*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-
> 

-- 
*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*
L.C. Karssen
Utrecht
The Netherlands

lennart at karssen.org
http://blog.karssen.org
GPG key ID: A88F554A
-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-*-

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 230 bytes
Desc: OpenPGP digital signature
URL: <http://lists.r-forge.r-project.org/pipermail/genabel-devel/attachments/20140218/a1f2c1e2/attachment-0001.sig>