On Wed, 27 Aug 2025 at 18:41, Oğuz <oguzismailuy...@gmail.com> wrote:
> Hmm. How much control do you have on the input format? If it's not too late,
> rather than swimming against the current maybe you should alter the input so
> it can be easily parsed by a POSIX shell
I have a lot of control, but I'm not sure what this would look like.
I could make the format say that the final line must contain just ".", and is
ignored. An EXIT trap could ensure that this final line is always output. But
now scripts which need to process each line separately have to ignore it, ie.
instead of:
while read; do
...
done
they need:
while read; do
if [ "$REPLY" = . ]; then
continue
fi
...
done
Some extra context is that some of the programs with this input/output protocol
are in other languages (usually when the processing involves using some
external library in that language), and they would also have to handle this
extra complexity in the protocol.
I could instead force some special way of representing a blank line, eg.
"<blank>", but then code which processes each word separately will need to be
more complex than the current:
for word in $line; do
...
done
unless the "special blank line" is "one or more spaces/tabs". But even then,
care needs to be taken to ensure that a space gets output instead of a pure
blank line.
It was around this point that I figured, maybe instead of fixing everything to
work around the shell quirk, I could just fix the shell quirk. :)
Kev
