I am attemting to build R from source and test R on NetBSD AMD64 and i386. I have run the configure and build steps successfully for R-2.3.1, R-2.6.0, and R-2.6.2. However, the suite of tests accessed by "make check" fails in numerous places. In some cases, the failure is expected, for example not using iconv libraries so the test for these libraries does not execute. In other cases, the errors are not as easy to diagnose. In one case it was an error with the function qcauchy (see error below). qcauchy(c(-Inf, 0), log = TRUE) == II is not all TRUE In addition: Warning message: NaNs produced in: qcauchy(p, location, scale, lower.tail, log.p) Execution halted My questions are general. Is it expected that the "make check" suite of tests will complete without errors? (R Administration/Installation page indicates that errors are not necessarily pathologic for a bad build). Is it necessary to sucessfully complete the "make check" suite of tests to "accept" an R installation for use? I work in a regulated environment and the initial thought was that "passing" the "make check" tests would be sufficient to accept a built-from-source R install. Any thoughts? Bill