🚀 go-pugleaf

Environment: 
gawk 2.14, patchlevel 0

Intel 386 running SYSVR3.2 from Interactive.
Sparc Sun running SUNOS 4.1x
DEC 5000-25 running ULTRIX 4.2

Problem:
There is no possibilty to use the NUL character (written as
^@ in the examples) in the FS variable as part of a regular
expression.  

For the tests below this input file was used:

tst.in:
------------------------------------------------------------
100 ^@ 200 ^@ 300
400  500 600
------------------------------------------------------------

Normaly this should work, but the string is cut off at the
NUL character.  The problem could be the usage of 'strlen'
in function 'set_FS' but I did not investigate further.
------------------------------------------------------------
gawk 'BEGIN {FS = "[ \t\0]+"}
{for (i = 1; i <= NF; i++) {
	print i, $i }}' <tst.in
------------------------------------------------------------
gawk: cmd. line:1: fatal: Premature end of regular expression: /[ 	/
------------------------------------------------------------

I tried this as a workaround, but this separates fields at
blanks, tabs, backslashes, and '0'
------------------------------------------------------------
gawk 'BEGIN {FS = "[ \t\\0]+"}
{for (i = 1; i <= NF; i++) {
	print i, $i }}' <tst.in
------------------------------------------------------------
1 1
2 ^@
3 2
4 ^@
5 3
6 
1 4
2 5
3 6
4 
------------------------------------------------------------

The NUL character is preserved in the string variable but
the effect remains the same.
------------------------------------------------------------
gawk 'BEGIN {tmp = "[ \t\0]+"; print tmp; FS = tmp}
{for (i = 1; i <= NF; i++) { 
	print i, $i }}' <tst.in
------------------------------------------------------------
[ 	^@]+
gawk: cmd. line:1: fatal: Premature end of regular expression: /[ 	/
------------------------------------------------------------

Peer Stritzinger
stritzi@physik.tu-muenchen.de