compiling

[rburghol]

See also:

• https://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html

Command:

• to suppress warnings, edit Makefile and set FFLAGS = -w
  • Note: this should only be used when isolating errors as warnings are actually useful info
• make jameswq 2>&1 >/dev/null | grep " " > make.errors
• ln -s /opt/model/james-chla/src/jameswq /usr/local/bin/jameswq

[rburghol]

First attempts.

• changed Makefile FC from FC = pgf90 to FC = gfortran
• complains of unknown options
  • gfortran has options:
  • https://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html
  • pgf90 options:
  • http://wiki.seas.harvard.edu/geos-chem/index.php/PGI_Fortran_compiler
• Command:
  • make jameswq 2>&1 >/dev/null | grep "" > make.errors

[rburghol]

• Error: Expecting variable in READ statement at (1)
  • Line is too long (more than 80 chars) https://gcc.gnu.org/legacy-ml/fortran/2011-02/msg00054.html

[rburghol]

• Many unrecognized command line option (see below).
• were ifort specific compiler things, ifort is no longer preferred in favor of gfortran
• eliminate all options FFLAGS=, though we may need to revisit

rob@deq2:/opt/model/james-chla/src$ make
gfortran  -Mpreprocess -fastsse -O2 -Mipa=fast -tp amd64  -mcmodel=medium  *.for  -o jameswq
gfortran: error: amd64: No such file or directory
gfortran: error: unrecognized command line option ‘-Mpreprocess’; did you mean ‘--preprocess’?
gfortran: error: unrecognized command line option ‘-fastsse’
gfortran: error: unrecognized command line option ‘-Mipa=fast’
gfortran: error: unrecognized command line option ‘-tp’; did you mean ‘-p’?
make: *** [Makefile:53: jameswq] Error 1
rob@deq2:/opt/model/james-chla/src$ https://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.htmlhttps://gcc.gnu.org/onlinedocs/gfortran/Preprocessing-Options.html

[rburghol]

Error: FORM specifier in OPEN statement at (1) has invalid value ‘binary’

• ifort specific OPEN statement, must change form='binary' to access='stream'
  • see: https://comp.lang.fortran.narkive.com/HlJxs3yF/alternative-to-ifort-s-form-binary
• Change form='binary' to form='unformatted'

[rburghol]

• access=append: this gives an error because we use "access=stream" to specify binary, so need to change this to position=append
• see: http://www-star.st-and.ac.uk/~pw31/CompAstro/lecture_6.pdf

Error: Duplicate ACCESS specification at (1)
zwqdump3.for:409:68:

  409 |    open(1,file='wqsed3d'//iyear//'.bin',access='stream',ACCESS='APPEND')
      |                                                                    1

[rburghol]

Some, but not all, OPEN statements fail with fom='unformatted' (fix put in for form='binary'):

• These lines are too long (> 72 chars)
• A continuation line command is needed, use "&" to indicate that the current line is a continuation of the previous line
• the "&" MUST be located at column 6 (i.e., 5 spaces, then the &) otherwise you will get a syntax error.

zwq3dinp.for:195:72:

  195 |         OPEN(555,FILE='Algaeavg.bin',STATUS='UNKNOWN',form='unformatted')
      |                                                                        1
Error: Syntax error in OPEN statement at (1)
make: *** [Makefile:54: jameswq] Error 1

[rburghol]

Make fails to find functions because they are in a source file with upper case FOR extension. Gives errors like this:

/usr/bin/ld: /tmp/ccgObvDi.o: in function `calmmt_':
calmmt.for:(.text+0xb8d4): undefined reference to `rvelplth_'

Rename source files:

for i in `ls *.FOR`; do
  newname=`echo $i | tr '[.FOR]' '[.for]'`
  mv $i $newname
done

[rburghol]

Getting close:

gfortran   -w  *.for  -o jameswq
/usr/bin/ld: /tmp/ccVjyjZW.o: in function `ssedtox_pcb_':
ssedtox_pcb.for:(.text+0xba94): undefined reference to `vmrmhp_'
/usr/bin/ld: /tmp/cc85DhzW.o: in function `timelog_':
timelog.for:(.text+0x1f): undefined reference to `time_'
collect2: error: ld returned 1 exit status
make: *** [Makefile:54: jameswq] Error 1

The above in ssedtox_pcb.for was another long line error, that somehow slipped past the compiler, i.e., it did not give a syntax error, but dropped the "," between vmrm and the first argument on the next line HP, (done in fortran function call format "inputs on one line, function rewritten outputs on the next line") and concatenated:

           CALL doboundary(Os,Asss,TEM(L,KC),SAL(L,KC),WINDST(L),umrm,vmrm,
        1   HP(L),DZC(KC),0,IWQKA)

Changed to:

           CALL doboundary(Os,Asss,TEM(L,KC),SAL(L,KC),WINDST(L),
     &       umrm,vmrm,
        1   HP(L),DZC(KC),0,IWQKA)

To fix the final timelog.for:(.text+0x1f): undefined reference totime_', I grepped for "subtroutine time" elsewhere, and found that all other calls were to functionTIMEF()`, so I just replaced it with that.