On Thu, March 27, 2008 3:47 pm, David Brown wrote: > On Thu, Mar 27, 2008 at 02:01:14PM -0700, Andrew Lentvorski wrote: > >> No, blame UNIX. UNIX is the wrong one is this instance. >> >> In human-computer interaction, anything which takes longer than .1 >> seconds >> causes the human brain to notice. At that point, status becomes a >> necessity. > > I think the time might be a little longer than 0.1 seconds, but definitely > less than 1 second. Git tried 2 seconds, and it was driving Linus batty > thinking that something had hung or crashed. Most git commands now print > out status after a few tenths of a second so you can tell something is > happening. > > One of my criteria for backup software is meaningful progress indication. > I'm using 'tar' now, and it goes through the 'buffer' program, which at > least prints the data transferred and the speed. > > Progress can be difficult to get right. A progress meter that quickly > moves to 99% and then stays there for a long time isn't very helpful, nor > is one that jumps backward periodically. There are certainly plenty of > those. > > Windows also has annoying animated progress indicators. People realize > they need to animate the progress indicator, but instead of tying it to > something that is actually making progress, they set a timer to animate > it. > So basically, it keeps running, even if everything has hung. Real useful > there. > > David
I completely disagree. Successful progress should be silent. If reassurance is necessary, a -v or -h (as in "print hash marks") can be added. Ask me how I feel about Y/N "should I proceed" questions that sit on my screen overnight. -- Lan Barnes SCM Analyst Linux Guy Tcl/Tk Enthusiast Biodiesel Brewer -- [email protected] http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-list
