I was baffled by why I wasn't getting much output from a somewhat
complicated SPECS stage, until I realized I had used NOWRITE and
forgotten the WRITE.  Then I was baffled by why I *did* get output at
EOF.  Here's a minimal example:

  pipe literal yes|literal no|specs nowrite select second 1-* 1|cons

I finally found my mistake by picking through the SPECS reference
again:  The NOWRITE is ignored in the run-out cycle because it applies
to the first reading station.  Putting it after the SELECT SECOND
instead makes it effective during the run-out cycle.

¬R

Reply via email to