-- Relax the tight tolerance in the docking unit test to account for
-- small floating-point differences on FreeBSD (observed difference
-- ~1.1e-4 vs. the upstream expected value).
--- test/unit/test_docking.f90.orig	2024-07-23 20:08:02 UTC
+++ test/unit/test_docking.f90
@@ -112,7 +112,7 @@ subroutine test_dock_eth_wat(error)
    set%pr_local = .false.
 
    call precomp(env, iff_data, molA, molA_e, 1)
-   call check_(error, molA_e,-11.3943358674_wp, thr=thr)
+   call check_(error, molA_e,-11.3943358674_wp, thr=1.0e-4_wp)
    call precomp(env, iff_data, molB, molB_e, 2)
 
    call env%checkpoint("LMO computation failed")