[flext] Compiling latest SVN OSX 10.5.8

[Stefano Papetti]

possibly in this regard, a student of mine (using Snow Leopard and gcc 
4.2.1) sent me his error log, which seems related to a gcc's issue only:

g++ -c -ffast-math -mmacosx-version-min=10.4 -O2   -fvisibility-inlines-hidden -arch ppc -mcpu=G4 -faltivec    -DFLEXT_USE_SIMD -DNDEBUG -DFLEXT_EXPORTS -DFLEXT_SYS=2 -DPD -I pd-darwin/release-single -I/Applications/Maths/Pd-extended.app/Contents/Resources/src source/flsimd.cpp -o pd-darwin/release-single/flsimd.opp_ppc
source/flsimd.cpp: In static member function ‘static void flext_single::AddSamples(t_sample*, const t_sample*, t_sample, int)’:
source/flsimd.cpp:1196: internal compiler error: Segmentation fault
Please submit a full bug report,
with preprocessed source if appropriate.
See<URL:http://developer.apple.com/bugreporter>  for instructions.
make[1]: *** [pd-darwin/release-single/flsimd.opp_ppc] Error 1
rm pd-darwin/release-single/flatom_pr.opp_ppc pd-darwin/release-single/flsupport.opp_i386 pd-darwin/release-single/flxlet.opp_i386 pd-darwin/release-single/fllib.opp_ppc pd-darwin/release-single/flbase.opp_i386 pd-darwin/release-single/flattr.opp_i386 pd-darwin/release-single/flext.opp_ppc pd-darwin/release-single/flext.opp_i386 pd-darwin/release-single/flattr_ed.opp_ppc pd-darwin/release-single/flthr.opp_i386 pd-darwin/release-single/flutil.opp_i386 pd-darwin/release-single/fldsp.opp_ppc pd-darwin/release-single/flatom_pr.opp_i386 pd-darwin/release-single/flatom.opp_ppc pd-darwin/release-single/flbuf.opp_ppc pd-darwin/release-single/fllib.opp_i386 pd-darwin/release-single/flsupport.opp_ppc pd-darwin/release-single/fltimer.opp_i386 pd-darwin/release-single/flxlet.opp_ppc pd-darwin/release-single/flbase.opp_ppc pd-darwin/release-single/flattr_ed.opp_i386 pd-darwin/release-single/flattr.opp_ppc pd-darwin/release-single/fldsp.opp_i386 pd-darwin/release-single/flthr.opp_ppc pd-darwin/release-single/fltimer.opp_ppc pd-darwin/release-single/flatom.opp_i386 pd-darwin/release-single/flutil.opp_ppc pd-darwin/release-single/flbuf.opp_i386
make: *** [build-release-single] Error 2

We solved just by disabling SIMD usage: this is not so important at the 
moment, since the student is only trying to get acquainted with flext, 
no critical development going on. Maybe we can try with the -Os flag.

Best,
S

Il 11/05/10 21:49, Arshia Cont ha scritto:
> Wierd because I tried switching to g++-4.2 with -O2 and compilation 
> failed... . But I did not really push on that.
>
> I believe, independent from gcc version, that 'symbols' included in 
> the library with -O2 causes conflicts with Max (and 
> undeterministically!). Or atleast this is what colleagues have told 
> me... . Not sure with Pd! Might be another misconception that has 
> lasted long... .
>
> Arshia Cont
>
>
>
>
> On May 11, 2010, at 5:24 PM, Thomas Grill wrote:
>
>> Hi Arshia,
>> i know that there have been problems documented, connected to 
>> compilation flags.
>> A combination that has always worked for me is g++-4.2 -O2 .
>> However, i considered g++-4.0.1 to be stable as well.
>> gr~~~
>>
>> Am 11.05.2010 um 14:54 schrieb Arshia Cont:
>>
>>> If like me, you are trying to compile Flext on OSX 10.5.8 for Max, 
>>> with gcc 4.0.1
>>>
>>> It is best to use -Os (instead of -O2) as the optimization options. 
>>> With -O2 many things will crap during object instantiation and also 
>>> destruction ... .
>>>
>>> (Thomas.. I think I solved all the stuff I bothered you for!). :)
>>>
>>> Arshia Cont
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> http://grrrr.org/ext/flext
>>>
>>> flext mailing list
>>> flext at grrrr.org
>>> http://grrrr.org/cgi-bin/mailman/listinfo/flext
>>
>>
>
> _______________________________________________
> http://grrrr.org/ext/flext
>
> flext mailing list
> flext at grrrr.org
> http://grrrr.org/cgi-bin/mailman/listinfo/flext
>