I just found a bug in fisher.test(). This is rw1080, on windows XP. A p-value can certainly not be Inf, but:
religion
Costumbres rel orig Religion Si Algunas veces Nunca católica 2121 4700 6234 prot/evan 100 216 2461 otra C 27 67 502 otra 0 0 14
fisher.test(religion, workspace=2000000)
Fisher's Exact Test for Count Data data: religion p-value = Inf alternative hypothesis: two.sided Kjetil Halvorsen