There isn't any technical reason for not checking. I just didn't think of
it when I first wrote the script. We are using links to create the new
index so we are not taking up any more disk space or inodes. But I suppose
something could still go wrong since the system has to create a new
directory entry.
Bill
On 6/20/07, Otis Gospodnetic <[EMAIL PROTECTED]> wrote:
Hi,
Looking at src/scripts/snapinstaller more closely, I saw this block of
code:
# install using hard links into temporary directory
# remove original index and then atomically copy new one into place
logMessage installing snapshot ${name}
cp -lr ${name}/ ${data_dir}/index.tmp$$
/bin/rm -rf ${data_dir}/index
mv -f ${data_dir}/index.tmp$$ ${data_dir}/index
Is there a technical reason why this wasn't written as:
logMessage installing snapshot ${name}
cp -lr ${name}/ ${data_dir}/index.tmp$$ && \
/bin/rm -rf ${data_dir}/index && \
mv -f ${data_dir}/index.tmp$$ ${data_dir}/index
This feels a little safer to me - I'd hate to have the main index rm
-rf-ed if the cp -lr command failed for some reason (e.g. disk full), but
maybe Bill Au & Co. have a good reason for not using &&'s. There may be
other places in various scripts that this might be applicable to, but this
is the first place I saw the extra safety possibility.
Thanks,
Otis
