Mark Kettenis wrote:
> I really think we should not encourage unportable code like that by
> giving an example in our manual page.
That's fair.
> I'm even tempted to say that you should leave the "-exec rm {} \;"
> example alone. The + here only works because rm(1) accepts multiple
> file arguments.
This whole adventure started because I was trying to delete 100000 old
emails. The default recipe, which execs one rm per file, was taking forever.
It would probably still be running. If I'm reading the find man page because
I'm trying to delete so many files I would overflow argv with shell wildcards,
I definitely want the turbo recipe.
> And I don't understand the sentence about recursively removing
> non-empty directories; rm(1) doesn't remove directories unless you use the
> -r option!
If some of the files you want to delete are directories, the obvious
solution when using xargs or -exec is to add -r to rm. But then this will
change behavior. I think it's possible to cook up a find invocation that uses
-type d -exec rmdir {} \; -o \! -type d -exec rm {} \; but as you see,
it gets unwieldy fast. I thought this would be worth highlighting.
