That's a slight misinterpretation (assuming I haven't misinterpreted what you wrote).
Writing a timestamp is a write operation on the file's metadata repository, not on the file itself. This is the inode entry in Unix filesystems, or some equivalent location in NTFS.
Updating mtime indicates a write operation on the file itself. As far as I can tell, this applies to all common filesystems.
oops66 wrote:
Case 1 - By default:
⊗ Preserve users data integrity (do "not touch my originals" policy ticked by default)...
Sorry, but I do not think this is part of the answer at all.
In my mind, the option of keeping original mtime is completely independent of an option that says whether image metadata is written to the image file.
This is not necessarily related to metadata/IPTC/XMP - it is purely related to file I/O and the options should therefore be located in a tab related to file I/O.
As far as I understand it, the sequence in a Unix/POSIX system is as follows:
1. XnView writes something to the image file.
2. the OS automatically updates the mtime timestamp to "now"
3. XnView then decides, based on user preference, whether to restore the original mtime value, probably by using something like utime().
The introduction of sidecar files means that presumably there are users who might want the same "keep original date" options applied to sidecar files.