On Fri, 2014-03-21 at 09:25 +0100, Lukas Middendorf wrote:
On 20/03/14 18:43, Patrick O'Callaghan wrote:
> I'm not saying it can do anything about it, I'm saying it should not be
> misleading. It claims to show transfer bandwidth and time to completion,
> but both are egregiously wrong. As I suggested earlier it should tell
> the user that these indicators are not reliable when the target is a
> remote server.
As I already wrote earlier, this is not limited to remote servers but
also applies to slow (removable) local storage (usb key drives, mobile
phones working as UMD, etc.) with the usually huge write caches you get
for current-day systems. It never is accurate when some kind of write
caching or buffering is involved, but for the named cases it is clearly
noticeable.
I agree, and I repeat that the user should have some indication of this.
Either that or just eliminate the indicator entirely if it's not giving
useful information (or rather reduce it to a simple flag that the
transfer hasn't completed yet).
poc