I have a very strange problem with chomp, but I can't produce a simple
test case that demonstrates it. I'm going to continue trying, but in
the meantime I thought I'd post what I _do_ know in the hope that
someone will say "ah yes..."

This is with the latest CVS.

I have an array, 'result', which I populate using 'readlines'. I then
pass it in to a routine which effectively does:


    for op in result
      if op =~ /^<<(\d+)>>/
        lineNo = $1.to_i
        op.gsub!(/^<<\d+>>/, '')
      end
      comment[lineNo] = op.chomp
      fullEscape(comment[lineNo], sp)
    end

Where fullEscape is:

  def fullEscape(line, space)
    $stderr.print "FE: ", line
    s = line.sub(/\s+$/, '').
      gsub(/\\/, "\\bs\?C-q").
      gsub(/([_\${}&%#])/, '\\\\\1').
      gsub(/\?C-q/, "{}").
      gsub(/\^/, "\\up{}").
      gsub(/~/, "\\sd{}").
      gsub("<<", "<{}<").
      gsub(/ /, space)
    $stderr.print "-> ", s, "\n"
    s
  end

Whn this runs in the the context of the larger program, I get the
following trace:


FE: [10, 15, 22, 120]-> [10,\ 15,\ 22,\ 120]^@
18: '[0.8333333333, 1.25, 1.833333333, 10.0]'
FE: [0.8333333333, 1.25, 1.833333333, 10.0]->
  [0.8333333333,\ 1.25,\ 1.833333333,\ 10.0]^@

Note the NULL character added to the end of the result.

However, if I change the op.chomp call in the loop to

   comment[lineNo] = op[0..-2]

The NULL goes away.

Similarly, if I change the call to fullEscape to be

   fullEscape(comment[lineNo]+'', sp)

the NULL goes away too.


The frustrating thing is that I haven't been able reproduce this in a
smaller test program.

Am I doing something stupid here?


Dave