Michał Górny wrote:
> + find "${D}" -type f -name '*.la' -print0 | while read -r -d '' f;
..
> + rm -f "${f}" || die
..
> + done
Don't pipe to read like that; it means the final command is in a subshell
and "die is /not/ guaranteed to work correctly if called from a subshell
environment."[1]
More seriously, the script doesn't actually get the correct filenames,
despite being written to handle any filename.
eg:
$ touch $' foo bar \n\t '
$ while read -r -d '' f; do echo "'$f'"; done < <(find . -type f -print0)
'./ foo bar'
You do it like this:
while read -rd ''; do
f=$REPLY;
..
done < <(find "$D" -type f -name '*.la' -print0)
eg:
$ while read -rd ''; do f=$REPLY; echo "'$f'"; done < <(find . -type f -
print0)
'./ foo bar
'
Or use: while IFS= read -rd '' f; do .. if you prefer.
See: help read # in a terminal.
It's called 'Process Substitution' if anyone wants to read about it in
man bash. The classic example with find is to get the list in an array:
arr=()
while read -rd ''; do
arr+=("$REPLY")
done < <(find "$dir" -type f .. -print0)
(perhaps conditionally though that's usually better done within find
which can later be handled on a per-file basis, or passed to:
foo "${arr[@]}"
..or if you just want to know whether there is a matching file:
if read -rd '' < <(find . -type f -print0); then
something matched
else nothing did
fi
They're both things I came up with a few years ago when I was learning
from #bash, which you are in dire need of, based on reading git-2.eclass.
[1] http://dev.gentoo.org/~ulm/pms/head/pms.html#x1-12600011.3.3
(11.3.3.6)
--
#friendly-coders -- We're friendly, but we're not /that/ friendly ;-)
