get check-ebcdic test running after merge

@IgorTodorovskiIBM i could use some help...

If I run the following:

#!/bin/sh
# Check of ICONV_EBCDIC_ZOS_UNIX environment variable.
set -e
set -x
host_os="$1"
iconv=../src/iconv_no_i18n

# This test is only meaningful when the EBCDIC encodings are included.
if true ; then

  if test "${host_os}" = 'openedition' ; then
    rm -f tmp-ok-lf tmp-ok-nel tmp-ok-x15 tmp-ok-x25
  fi

  sh -c "printf '\150\145\154\154\157\012' > tmp-ok-lf"
  sh -c "printf '\150\145\154\154\157\302\205' > tmp-ok-nel"

  sh -c "printf '\210\205\223\223\226\025' > tmp-ok-x15"
  sh -c "printf '\210\205\223\223\226\045' > tmp-ok-x25"

  if test "${host_os}" = 'openedition' ; then
    # On z/OS, make all the files 'untagged'
    chtag -r tmp-ok-lf
    chtag -r tmp-ok-nel
    chtag -r tmp-ok-x15
    chtag -r tmp-ok-x25
  fi

  # Check that by default, EBCDIC 0x15 maps to U+0085.
  export ICONV_EBCDIC_ZOS_UNIX=JUNK
  unset ICONV_EBCDIC_ZOS_UNIX

  $iconv -f UTF-8 -t IBM-1047 < tmp-ok-nel > tmp-out1
  $iconv -f ASCII -t IBM-1047 < tmp-ok-lf > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-x15
  /bin/cmp tmp-out2 tmp-ok-x25

  $iconv -f IBM-1047 -t UTF-8 < tmp-ok-x15 > tmp-out1
  $iconv -f IBM-1047 -t ASCII < tmp-ok-x25 > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-nel
  /bin/cmp tmp-out2 tmp-ok-lf

  # Check that with the ZOS_UNIX surface, EBCDIC 0x15 maps to U+000A.

  $iconv -f ASCII -t IBM-1047/ZOS_UNIX < tmp-ok-lf > tmp-out1
  $iconv -f UTF-8 -t IBM-1047/ZOS_UNIX < tmp-ok-nel > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-x15
  /bin/cmp tmp-out2 tmp-ok-x25

  $iconv -f IBM-1047/ZOS_UNIX -t ASCII < tmp-ok-x15 > tmp-out1
  $iconv -f IBM-1047/ZOS_UNIX -t UTF-8 < tmp-ok-x25 > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-lf
  /bin/cmp tmp-out2 tmp-ok-nel

  # Check that the ZOS_UNIX surface can be used with //TRANSLIT and //IGNORE.

  sh -c "printf '\357\275\210\342\204\257\360\235\232\225\360\235\232\225\342\200\242\012'" \
    | $iconv -f UTF-8 -t IBM-1047/ZOS_UNIX/TRANSLIT > tmp-out1
  /bin/cmp tmp-out1 tmp-ok-x15

  sh -c "printf '\150\145\154\154\157\342\202\254\012'" \
    | $iconv -f UTF-8 -t IBM-1047/ZOS_UNIX/IGNORE > tmp-out1
  /bin/cmp tmp-out1 tmp-ok-x15

  # Check that with a specific environment variable, EBCDIC 0x15 maps to U+000A.
  ICONV_EBCDIC_ZOS_UNIX=1
  export ICONV_EBCDIC_ZOS_UNIX

  $iconv -f ASCII -t IBM-1047 < tmp-ok-lf > tmp-out1
  $iconv -f UTF-8 -t IBM-1047 < tmp-ok-nel > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-x15
  /bin/cmp tmp-out2 tmp-ok-x25

  $iconv -f IBM-1047 -t ASCII < tmp-ok-x15 > tmp-out1
  $iconv -f IBM-1047 -t UTF-8 < tmp-ok-x25 > tmp-out2
  /bin/cmp tmp-out1 tmp-ok-lf
  /bin/cmp tmp-out2 tmp-ok-nel

  rm -f tmp-in* tmp-out* tmp-ok*
fi

exit 0

under 'sh' as:

./check-ebcdic_EBCDIC openedition

then it runs clean. Trying to run this under bash when the file is encoded in ASCII is causing me grief. It would be good to see if there is a way to a) generate files from hex or octal bytes where no translation occurs and b) prevent all tagging.

Open to ideas of the best way to fix this up. This came originally from: check-ebcdic in the tests directory

ZOSOpenTools / libiconvport

get check-ebcdic test running after merge #18