[PATCH 3/3] kernel32: Added support for MOVEFILE_WRITE_THROUGH in MoveFileWithProgress

[George Gonzalez]

Ah, righto!

One interesting thing from that link,  it discusses whether sync() 
actually writes anything out physically, and whether it guarantees that 
it's physically on the disk, physically updated in the directory, or 
maybe one or the other, and whether those are atomic, and is it at the 
drive, or actually written from the drive internal cache to the 
permanent parts of the drive, and how does all that relate to network 
files?  So many possible ways to weasel out of a definite and firm sync.


On 02/07/2018 02:53 PM, Matteo Bruni wrote:
> 2018-02-07 20:39 GMT+01:00 George Gonzalez <ggonzalez at codeweavers.com>:
>> Maybe fsync()?
> The point is that you want to use a Win32 API instead of a POSIX one.