From kde-core-devel  Mon Nov 25 23:35:18 2013
From: Alexander Neundorf <neundorf () kde ! org>
Date: Mon, 25 Nov 2013 23:35:18 +0000
To: kde-core-devel
Subject: Re: KMountPoint::probablySlow and cifs mount points
Message-Id: <201311260035.18639.neundorf () kde ! org>
X-MARC-Message: https://marc.info/?l=kde-core-devel&m=138542237312558

On Tuesday 26 November 2013, Thomas Lübking wrote:
> On Montag, 25. November 2013 23:31:59 CEST, Alexander Neundorf wrote:
> > hmm, doesn't have to be really complicated, just a little bit more
> > sophisticated than it is now, e.g. like the following, which is
> > not too hard
> > to do:
> > probablySlow = (NFS || CIFS) && (host is not in my own subnet || NIC is
> > wireless)
> 
> - wired NIC connected to WLAN router/repeater
> - wired 10Mbps NIC

Serious question: do systems, which run KDE4 and are connected to a 10 Mbps 
wired network exist ?

> - remote system on WLAN
> - remote system reads from USBv1.1 or CDROM

...or local system reads from USBv1.1 or CDROM

> - traffic shaping
> - SMB/CIFS is slow even on fast connections (see the bug Albert linked)

Fast... if I see it correctly, this is a high-bandwith, but long latency 
connection. Are those machines all in the same subnet ?
 
> OTOH, NFS to another subnet (the next in-house router, GBit) can be pretty
> fast. For servers, read & write can largely differ and whatnot.

Yes, distributed network file systems can even be faster than local ones.

I think we agree that there are enough corner cases, and that it is hard to 
find a really correct way.
But, can we still improve the logic a bit ?

Alex