Julian Sikorski wrote: > Hi, > > I am trying to make gnome-chemistry-utils ready for the mass rebuild. > Unfortunately, there are some issues which are beyond my skills. The > first one is the fact that some adjustments for new xulrunner are > needed. With gcc-4.4, however, build fails even earlier: > > element.cc: In function 'void ReadValue(const char*, GcuValue&)': > element.cc:49: error: invalid conversion from 'const char*' to 'char*' > element.cc: In function 'void ReadDimensionalValue(const char*, > GcuDimensionalValue&)': > element.cc:58: error: invalid conversion from 'const char*' to 'char*' > > Full build log is here: > http://belegdol.fedorapeople.org/build.log > > Upstream unfortunately cannot reproduce the problem, here is his comment: > > > Looks like the issue comes with strchr. > > man strchr says: > char *strchr(const char *s, int c); > > I don't see what is wrong since the argument is a const char* string. > > I can't reproduce the issue. May be the signature of strchr changed in > glibc? > Anyway, the above signature is not consistent since it returns a char* which > is a substring of a const char*. I agree, that's not your fault. It's what the standard says it has to be. > If this is the case, there is a need for a test in configure to find which > signature version is available. I don't see why. It looks to me like you can simply do this: static void ReadDimensionalValue (char const *source, GcuDimensionalValue &value) { const char *buf, *dot; char *tmp; value.value = strtod (source, &tmp); buf = tmp; dot = strchr (source, '.'); value.prec = (dot)? buf - dot - 1: 0; value.delta = (*buf == '(')? strtol (buf + 1, NULL, 10): 0; } Slightly weird, but it'll do. Andrew. -- fedora-devel-list mailing list fedora-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/fedora-devel-list
