Hi Dirk,
On Sun, 11 Jun 2006, Dirk Reiners wrote:
> On Thu, 2006-05-11 at 21:29 +0200, Akos Balazs wrote:
>>
>> and run into some "usual" inlining problems, so I had to patch
>> CommonConf/common.x86_64-unknown-linux-gnu-g++.in (see attachment).
>> I've checked the CVS web interface at SF, and it seems this is not yet
>> in. I've also noticed that Dirk removed the `-m64' flag from LANG_FLAGS
>> (so the patch may not apply cleanly to the current CVS) is there a
>> reason? I believe it's necessary to have gcc compile code as 64 bit,
>> isn't it?
>
> On a 64 bit system it creates 64 bit code by default, so you don't need
> the flag.
OK. What about the patches related to proper inlining? It seems you only
have it for gcc 4.x whereas later gcc 3.x versions also require them
AFAICS. This probably also goes for 32 bit systems, but I haven't tried
compiling with gcc on such systems for a while.
>> Another small issue: since I compile on a dualcore system, I figured I
>> could save some time by running make as `make -j2 opt' etc., so that I
>> compile two files at the same time. The compile ran fine, but the
>> Window* libs (e.g. WindowGLUTLib) didn't link (and I think didn't
>> compile completely either.) I had to manually enter
>> `Builds/x86_64-unknown-linux-gnu-g++/WindowGLUTLib/' and call `make'
>> from there, has anyone seen such behaviour before? Without the `-j2'
>> switch it seems to compile/link everything correctly. Is it a
>> dependency problem exposed by the parallel make, or am I missing
>> something as usual? :)
>
> Hm, haven't seen that before. But '-j2' in general doesn't work anyway
> (-j doesn't go into subdirectories), you have to use setenv OSGSUBPARJBS
> 2 for that.
It seemed to work more or less (just missing some libs), but Gerrit
already answered this one. :-) (The keyword being OSGSUBPARJOBS of course.
:)
Thanks,
Akos
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users
