I am having difficulty compiling my postgresql 7.2.3 on a CentOS 5. The same code works on our build machine which is running RedHat 7.2.
I have checked many of the configure(1) output with no major diff. Capturing the output of make reveals a silly problem which I am failing to find the root of. The problem is:
Compilation moves forward and at one point it does a "make -C libpq all". This operation builds all the object files and even ar(1) them to create libpq.a. But next step, attempting to run ranlib(1) fails because the ranlib command was misconstructed as in
Note how "ranlibCFLAGS" should've been "ranlib CFLAGS" So it looks like in my CentOS 5 something has changes such that the variable substituions are wrong. Which points to things that are autogenerated, which points to configure(1).
Any tips on the actual problem or how to get to the bottom of it is appreciated. Perhaps a few make tricks would surface the problem.