[lmi] Unused but set variables

[Vadim Zeitlin]

 Hello,

 After the CI was triggered by a recent commit to master removing
gui_test.sh, there were two breakages: one was expected, as the script was
used for running the GUI tests in the MinGW build (what wasn't expected is
that it's still broken, after replacing the script with the manual
invocation of wx_test, but this is a different topic), while another one
was not, as it was due to the clang version change in Debian Sid which
happened in the meanwhile. The new clang version, 13, has brought (at
least) two new warnings: one is about not declaring a copy ctor when the
assignment operator is declared (even if it's only declared as deleted,
which seems a bit stupid) and happens only in wx headers and so only
affects the autotools build which doesn't use -isystem for them.

 But the other new warning, "-Wunused-but-set-variable", does happen in lmi
code itself and both of its occurrences seem to indicate actual problems:

 The first one is

group_values.cpp:366:20: error: variable 'eoy_inforce_lives' set but not used 
[-Werror,-Wunused-but-set-variable] 

and, indeed, eoy_inforce_lives is assigned to, but never read from. I'm not
sure what was meant here, but it seems suspicious that we go to the trouble
of calling InforceLivesEoy() for each cell, yet never use the result. But
if this is indeed the right thing to do, could we remove eoy_inforce_lives
entirely?

 The second one is

interest_rates.cpp:719:23: error: parameter 'AnnualSepAcctIMFRate' set but not 
used [-Werror,-Wunused-but-set-parameter]

and is even weirder, as we change a function parameter, which is curious on
its own, and not only we don't use it later, but we don't use it at all,
which is even curiosier. I have no idea about what should really happen
here, but the simplest change avoiding the warning would be to simply
remove AnnualSepAcctIMFRate parameter from DynamicMlySepAcctRate. I know
that this function is supposed to be redesigned entirely (there is a
comment about it), but it's probably not going to happen in the immediate
future, while I'd like to make clang CI build pass again reasonably soon
and I think it would be better to avoid disabling this warning, as it seems
quite useful -- at least judging from two true positives above.

 Please let me know if you think we can apply my proposed fixes above or if
you'd prefer to do something more radical. Or if you'd rather just disable
the warning, after all.

 Thanks in advance!
VZ

pgpsJFOapwsmX.pgp
Description: PGP signature