Re: psu-lush 1.2
Ralf Juengling <[email protected]> Mon, 10 Dec 2007 10:10:13 -0800 (PST)
| Newsgroups | gmane.lisp.lush.devel |
|---|---|
| Message-ID | <[email protected]> |
Thanks Raymond, I was caught up with stuff the last two weeks. On Fri, 30 Nov 2007, Raymond Martin wrote: > during compile this message: > > gcc -std=c99 -DHAVE_CONFIG_H -DNO_DEBUG -Wall -O3 -march=i686 -mmmx -msse -I../include -pthread -I/usr/include/freetype2 -c index.c > index.c:529: warning: C99 inline functions are not supported; using GNU89 > index.c:529: warning: to disable this warning use -fgnu89-inline or the gnu_inline function attribute .. I don't understand this warning. It says -std=c99 on the command line, so gcc should expect c99 code. Is this an older version of the compiler? > During start up of psu-lush and loading of modules > > [iterator-classes.lsh] > > *** module-load : dld/bfd error > *** File in wrong format : "/mnt/hdb11/raymond/work/science/compsci/languages/l ... > > Can't see which file it is referring to, but it starts with iterator-classes.lsh > > Did you see these problems already? The second problem might only be on my computer > because I am having problems with flaky hardware that locks up X when I compile anything > intense. Yes, I found that problem too late, unfortunately. It may occur (only) when you do a system install and 'varlushdir' is defined. iterator-classes is a module that is imported by sys/stdenv.lsh and so all definitions in there are in the image file (the .dump file). iterator-classes contains build instructions (dhc-make) and so there is a dependency on the corresponding object file in the image file. Now the problem is this: On interpreter startup this object file gets loaded from the lushdir directory. This object file may look outdated (after copying by make install) and the next time a line (libload "iterator-classes") is read, iterator-classes is read again. The interpreter compiles again, now in the varlushdir location. Since the old object file is not in the varlushdir location, the interpreter does not unload the old object file (as it should), resulting in the above error. I'm still thinking about the right way to fix this. The problem would not occur if 'make install' instructions would copy the time information along with the files. This is desirable, I think, because it would enable making use of a pre-compiled standard library. That way you don't only save compilation time but also space in varlushdir (usually in you home directory). Ralf ------------------------------------------------------------------------- SF.Net email is sponsored by: Check out the new SourceForge.net Marketplace. It's the best place to buy or sell services for just about anything Open Source. http://sourceforge.net/services/buy/index.php