DABS vs ABS conflict for polarized processes

Hi folks,

Unsure if this is a bug but I have observed anecdotally that there is a ABS vs DABS conflict when trying to simulate polarized processes. I get things like:

gfortran -w -fPIC -O -fbounds-check -ffixed-line-length-132 -c -o MP_FFV8P1L_3.o MP_FFV8P1L_3.f
     MP_FFV8P1L_3.f:29:35:

        29 | IF (ABS(P3(0))*1E-10.GT.DABS(P3(1))) P3(1)=1D-30
           | 1
     Error: ‘a’ argument of ‘dabs’ intrinsic at (1) must be REAL
     MP_FFV8P1L_3.f:30:35:

        30 | IF (ABS(P3(0))*1E-10.GT.DABS(P3(2))) P3(2)=0D0
           | 1
     Error: ‘a’ argument of ‘dabs’ intrinsic at (1) must be REAL
     MP_FFV8P1L_3.f:31:35:

        31 | IF (ABS(P3(0))*1E-10.GT.DABS(P3(3))) P3(3)=1D-15
           | 1
     Error: ‘a’ argument of ‘dabs’ intrinsic at (1) must be REAL

My workaround is to modify ./aloha/aloha_writers.py line 640 (in 3.4.0.bzr and 2.9.16.gitLTS) from

out.write(" IF (DABS(%(P)s(0))*1e-10.gt.DABS(%(P)s(%(i)s))) %(P)s(%(i)s)=%(val)s\n"

to

out.write(" IF (ABS(%(P)s(0))*1e-10.gt.ABS(%(P)s(%(i)s))) %(P)s(%(i)s)=%(val)s\n"

That is, replace DABS with ABS. Things work after that. (Of course, I need to regenerate and re-output my process.) However, I am not sure if this is symptomatic of a larger REAL(4) vs REAL(8) mismatch somewhere.

If anyone needs it, I have a non-public UFO and script that can reproduce the issue.

cheers,
richard