[GEM-dev] configure (again): call for testing....

[james tittle]

On Jul 7, 2005, at 7:17 AM, IOhannes m zmoelnig wrote:

> james tittle wrote:
>
>> ok,
>> ...just some thoughts:
>> *  -framework isn't a CXXFLAG, it's a linker flag...
>>
>
> ah yes, should've read this mail before i answered the last one.

...fixed!

>> *  probably should default to no-x11 on osx, with the option of  
>> re- configuring if ya want to try it with gem (never have  
>> myself):  plus,  the libs will probably cause havoc conflicting  
>> with the frameworks  (especially -lGL)
>>
>
> hmm, günter, what is the suggested way to handle such things with  
> configure ?

...fixed, but I'm still getting some x11 stuff because I have  
imagemagick via fink...unfortunately, --without-imagemagick doesn't  
seem to work, but --without-ImageMagick does:  is it possible to make  
this work both ways?

>> *  m_pd.h is found for me, but then I don't install into /usr/local
>>
>
> /usr/local is just the default installation path of configure; you  
> should be able to set it with a flag to configure (but it is only  
> of importance, if you want to install Gem via "make install";  
> otherwise you are safe by ignoring it)


...apparently here I made a typo!  I meant to say that m_pd.h IS NOT  
being found, even with "--includedir=/absolute/path/to/pd/src":  how  
else can I set this (outside of PATH environment variables)?


>> *  "-shared -Wl,-export-dynamic" isn't the correct way to link:   
>> we  use "-bundle -bundle_loader /where/is/pd"
>>
> well, again: configure should figure out, that "-shared  ..." is  
> invalid. for bundle-loader i have obviously used the wrong options.

...haven't gotten this far, but it is good to have the "--with-pd=/ 
path/to/pd/bin" that you added...

>> *  I'm not sure if we should use -O3,s is seemingly recommended  
>> in  apple's docs (basically -O2 + smaller size)...
>>
>
> aha.
> reading the man-page for g++ it says:
> "If you use multiple -O options, with or without level numbers, the  
> last such option is the one that is effective."
>
> and on "-Os" it says that it turns _off_ optimizazions that might  
> increase size.
>
> so to be on the safe side, it would probably be good to use "-Os - 
> O3" (?)

...that's the thing:  why have two "-Ox" flags if only the last one  
is effective?  As far as -Os goes, apparently size increasing  
optimizations aren't necessarily desirable on a ppc:  apple docs  
always suggest using -Os...otoh, I've never done profiling of -Os vs.  
-O3 code to see if there's any real difference in our case, so it's  
just speculation...

> i just noticed that our G4 is online and did some testing myself  
> with the new configure thing.
> i hopefully fixed a lot of issues you 2 were having.
> i'll check them in in a second.
> finally it compiled, linked and stripped cleanly.
> unfortunately, i was having a break during compilation and when i  
> cam back from dinner, the G4 was turned off. (it is physically  
> located in one of our studios, and i guess it was too loud for the  
> people working in there)
> so i don't know whether pd is able to load the produced library  
> (and of course i have no idea, whether Gem will work once it is  
> loaded)

...what, ya'll don't have it in an anechoic chamber? ;-)

> ah, one last thing:
> i have also moved the content of macos_math.{h,cpp} into Base/ 
> GemMath.h.
> so in future we would just have to include "Base/GemMath.h" as  
> opposed to <math.h> and the float vs. double thing will get  
> corrected automatically.

...thanx for doing the janitorial cleanup!

jamie