On 02/20/2014 08:39 PM, ronf wrote: > Andrew Haley wrote >> On 02/13/2014 11:34 PM, ronf wrote: >>> I was surprised that the procedure for building the glibc packages (from >>> source) relied so heavily on glibc packages that were already installed. >>> One >>> great feature of a compiler collection like GCC is the ability of >>> compilers >>> to compile themselves from source. I was surprised that the build process >>> was so dependent on kernel libraries. I can understand gcc using kernel >>> libraries for its own internal operations, but I don't believe the build >>> of >>> a GCC source package should be including kernel source (headers) from >>> outside the package or needing to link the compiled object code with >>> kernel >>> libraries other than those generated by part of the build. >> >> What kernel libraries are you talking about? Please be specific. > > I am sorry I didn't check my messages for so long. Since GCC now builds > easily, I don't see the errror messages naming the libraries and, thus, I > don't know what libraries were missing. However, I do remember the procedure > I used. > > First, since stubs-32.h was missing I copied stubs-64.h to stubs-32.h in an > attempt to trick the build into continuing. (This was a mistake since it was > unnecessary and wrong.) > > Second, I configured using the --enable-multilib option (another mistake). > > Third, I entered "make" and got error messages regarding missing libraries. > > I hope you will tell me there were no missing kernel libraries but just > confusion caused by my unusual bad procedure. I hope GCC really does compile > itself from source code without needing binary libraries other than > libraries used only for the internal operations of the compiler used to > build GCC. It really does. Of course, it needs a C library. Andrew.
