From: "Ben Collins-Sussman" <[EMAIL PROTECTED]> Sent: Tuesday, May 22, 2001 8:44 AM
> OK, it's not actually a patch, because I'm not sure where it ought to > live. I've attached it below. > > The function apr_dir_remove() won't work on a non-empty directory, so > this routine fills the gap. It's equivalent to 'rm -rf'. My personal preference is that the remove continues to run, leaving only droppings it can't remove. Nothing irks me more than fixing the 'one file' that couldn't be removed, only to trip over the next one on the next pass. And save the first failure result for return to the caller. Does it make sense to apply some lstat check against the tree, such that symlinks' targets aren't blown away? I'm not clear if your patch protects that or not. Bill
