On Mon, Sep 14, 2015 at 4:58 PM, Eric Covener <[email protected]> wrote: > On Mon, Sep 14, 2015 at 10:52 AM, Yann Ylavic <[email protected]> wrote: >> Oh I see, I guess FILE_FLAG_DELETE_ON_CLOSE is not suitable there >> since files are opened/closed by different processes... >> But if I read DeleteFile() doc correctly, for non-mapped files it >> should work (delayed until the last handle is closed), or really >> DeleteFile() is not documented correctly! > > > I vaguely recall conflicting info in that area from the > docs/forums/comments. But I know without a restart, the closure > within the same process leaves an un-deletable file.
Hm, that may be a good news for the slotmem_shm case where we always (try to )remove the file before the SHM is destroyed, so that's in the same process. I already tried to invert both for PR 58024, but it was before introducing the generation number. So it may work with the full monty (already proposed to the OP, wait and see).
