This only triggers when running maxima-sbcl on 32 bit, which gives:

sage: maxima('exp(-sqrt(x))').nintegral('x',0,1)
(0.5284822353142306, 4.163291933423352e-11, 231, 0)

Compare to the "normal" result (64 bit, maxima-ecl on 32 bit):

sage: maxima('exp(-sqrt(x))').nintegral('x',0,1)
(0.5284822353142306, 4.163314137883845e-11, 231, 0)

Since the difference is in the bound for the error in the numerical integration, it doesn't seem very important.

Component: numerical

Author: Gonzalo Tornaría

Branch/Commit: 30259c6

Reviewer: Frédéric Chapoton

Issue created by migration from https://trac.sagemath.org/ticket/34669