@sneumann can you confirm this? If same for you, I think this is a historic documentation error as @meowcat handled this differently in RMassBank.
https://github.com/MassBank/RMassBank/blob/master/R/leCsvAccess.R#L315
Then to solve this we need to fix docs. This was still on my old and working rcdk3.4.3 (about to test on new one)

#this DOESN'T WORK
smiles <- c("c1ccccc1", "CCC")
m <- parse.smiles(smiles[1])[[1]]
do.aromaticity(m)
do.typing(m)
do.isotopes(m)
convert.implicit.to.explicit(m)
get.mol2formula(m)
get.exact.mass(m) #<= THIS FAILS - and ONLY THIS!
get.natural.mass(m)

m <- parse.smiles(smiles[2])[[1]]
do.aromaticity(m)
do.typing(m)
do.isotopes(m)
convert.implicit.to.explicit(m)
get.mol2formula(m)
get.exact.mass(m) #<= THIS ALSO FAILS! With non-aromatic SMILES
get.natural.mass(m)

#This now works
m <- parse.smiles(smiles[2])[[1]]
do.aromaticity(m)
convert.implicit.to.explicit(m)
do.aromaticity(m) #<= NOTE THE ORDER! THE FUNCTION STILL RETURNS FALSE BUT ... 
do.typing(m)
do.isotopes(m)
get.mol2formula(m)
get.exact.mass(m) #<= THIS NOW WORKS!
get.natural.mass(m)

#This worked all along:
library(RMassBank)
mol <- getMolecule(smiles[1])
get.exact.mass(mol)

with latest rcdk this order dependency is gone.
other attached packages:
[1] rcdk_3.4.9.1 rcdklibs_2.2 rJava_0.9-8 Rcpp_0.12.17

if I run the "broken" code from above:

smiles <- c("c1ccccc1", "CCC")
m <- parse.smiles(smiles[1])[[1]]
do.aromaticity(m)
do.typing(m)
do.isotopes(m)
convert.implicit.to.explicit(m)
get.mol2formula(m)
get.exact.mass(m) #<= THIS FAILS - and ONLY THIS!
get.natural.mass(m)


m <- parse.smiles(smiles[2])[[1]]
do.aromaticity(m)
do.typing(m)
do.isotopes(m)
convert.implicit.to.explicit(m)
get.mol2formula(m)
get.exact.mass(m) #<= THIS ALSO FAILS! With non-aromatic SMILES
get.natural.mass(m)

then the functions that failed above now return correct values.

> mol <- getMolecule(smiles[1])
> get.exact.mass(mol)
[1] 78.04695
> smiles <- c("c1ccccc1", "CCC")
> m <- parse.smiles(smiles[1])[[1]]
> do.aromaticity(m)
[1] TRUE
> do.typing(m)
> do.isotopes(m)
> convert.implicit.to.explicit(m)
> get.mol2formula(m)
cdkFormula:  C6H6 , mass =  78.04695 , charge =  0 
> get.exact.mass(m) #<= THIS FAILS - and ONLY THIS!
[1] 78.04695
> get.natural.mass(m)
[1] 78.11206
> 
> 
> m <- parse.smiles(smiles[2])[[1]]
> do.aromaticity(m)
[1] FALSE
> do.typing(m)
> do.isotopes(m)
> convert.implicit.to.explicit(m)
> get.mol2formula(m)
cdkFormula:  C3H8 , mass =  44.0626 , charge =  0 
> get.exact.mass(m) #<= THIS ALSO FAILS! With non-aromatic SMILES
[1] 44.0626
> get.natural.mass(m)
[1] 44.09573

@sneumann can you confirm? Seems that would solve this issue for now?

For the record from John:

MolecularFormulaManipulator
getTotalExactMass(mf); => for each isotope adds up all the exact mass, if an isotope is not specified (null) the major isotope will be used
getNaturalExactMass(mf); => for each isotope adds up the natural mass, ignoring any specified isotopes.
getTotalMassNumber(mf); => adds up the major isotope mass number of each isotope element, ignoring any specified isotope
getMajorIsotopeMass(mf); => adds up the major isotope mass of each isotope element, ignoring any specified isotope

AtomContainerManipulator
getTotalExactMass(IAtomContainer); => for each atom adds up all the exact mass, if isotopes are not specified you'll get a NPE. ImplH count is multiplied with 1H mass
getNaturalExactMass(IAtomContainer); => for each atom adds up the natural mass, ignoring any specified isotopes. ImplH count is multiplied with H natural mass.
getMolecularWeight(IAtomContainer); => for each atom adds the exact mass if specified, or the natural mass if unspecified (null). ImplH count is multiplied with H natural mass.

Adding bits & pieces from email conversations here for the record while we look into this:

Natural Mass are from BODR - https://github.com/cdk/cdk-build-util/blob/master/src/main/resources/net/sf/cdk/tools/bodr/chemicalElements.xml
and BODR=Blue Obelisk Data Repository - https://sourceforge.net/projects/bodr/
and natural mass == average mass == natural abundance? [to check with Chris/Egon]
Data from IUPAC 2009:
https://github.com/cdk/cdk/blob/cdk-1.4.x/src/main/org/openscience/cdk/config/data/isotopes.xml

and think about:
get.mass(mol, type = c('total.exact', 'natural.exact', 'mol.weight'))
rather than 3 separate methods. You could even combine this with the two methods that take a MF as the first argument

Further bits:

> m <- parse.smiles("CCNC")[[1]]
> do.aromaticity(m)
[1] FALSE
> do.typing(m)
> do.isotopes(m)
> 
> get.exact.mass(m)
[1] 59.0735
> m <- parse.smiles("CCNC")[[1]]
> do.aromaticity(m)
[1] FALSE
> do.typing(m)
> do.isotopes(m)
> convert.implicit.to.explicit(m)
> get.exact.mass(m)
[1] "Java-Object{java.lang.NullPointerException}"
Error in get.exact.mass(m) : 
  Couldn't get exact mass. Maybe you have not performed aromaticity, atom type or isotope configuration?
> m <- parse.smiles("CCNC")[[1]]
> get.exact.mass(m)
[1] "Java-Object{java.lang.NullPointerException}"
Error in get.exact.mass(m) : 
  Couldn't get exact mass. Maybe you have not performed aromaticity, atom type or isotope configuration?

other attached packages:
[1] RUnit_0.4.31    rcdk_3.4.3      rcdklibs_2.0    rJava_0.9-10    devtools_1.13.3 RMassBank_2.9.1 Rcpp_0.12.19