On Fri, 10 Feb 2017 12:07 am, eryk sun wrote: > On Thu, Feb 9, 2017 at 11:46 AM, Steve D'Aprano > <steve+pyt...@pearwood.info> wrote: >> >> So to summarise, os.rename(source, destination): >> >> - is atomic on POSIX systems, if source and destination are both on the >> same file system; >> - may not be atomic on Windows? >> - may over-write an existing destination on POSIX systems, but not on >> Windows; >> - and it doesn't work across file systems. > > On Windows in 2.7 and prior to 3.3, os.rename will silently copy and > delete when the destination isn't on the same volume.
Will the copy overwrite an existing file? > It may even > silently leave the original file in place in some cases -- e.g. when > the file is read-only and the user isn't allowed to modify the file > attributes. > > If the destination is on the same volume, renaming should be atomic > via the system calls NtOpenFile and NtSetInformationFile. Ultimately > it depends on the file system implementation of > IRP_MJ_SET_INFORMATION, FileRenameInformation [1]. > >> The link/unlink trick: >> - avoids over-writing existing files on POSIX systems at least; >> - but maybe not Windows? > > This works for renaming files on Windows as long as the file system > supports hard links (e.g. NTFS). It's not documented on MSDN, but > WinAPI CreateHardLink is implemented by calling NtSetInformationFile > to set the FileLinkInformation, with ReplaceIfExists set to FALSE, so > it fails if the destination exists. Note that this does not allow > renaming directories. See the note for FileLinkInformation [1]; NTFS > doesn't allow directory hard links. But why bother with this 'trick' > on Windows? I don't know, that's why I'm asking for guidance. I don't have a Windows system to test on. On Windows, how would you implement a file rename (potentially across file system boundaries) which will not overwrite existing files? Just by calling os.rename()? -- Steve “Cheer up,” they said, “things could be worse.” So I cheered up, and sure enough, things got worse. -- https://mail.python.org/mailman/listinfo/python-list
