Another data point on delays in focus update.
A month or so ago, I posted about some slow focus updates I was seeing in xmonad. http://article.gmane.org/gmane.comp.lang.haskell.xmonad/12074 I noticed another correlation when this is happening, and thought I'd jot it down in this forum, in the hopes that it eventually contributes to a solution. During the period when the focus is slowly traversing a lot of intervening windows, I've found that I peg a CPU running the X server; My Xorg process goes to 99%CPU and stays there for the entirety of the delay. I've been watching the X cpu behavior because I've also noticed that some Firefox browser behavior will similarly peg the Xorg process for seconds at a time, and I was trying to suss that out. I'm kind of free-associating here, but I'm wondering if xmonad's use of X resources is leaving Xorg with a long list of [wumpus] that the X server can't collect or collapse or something, I have no idea what the wumpus might be, but... - Allen S. Rout - Climbing back under his rock
Does it happen with an out of the box xmonad ? With no user config?
On Tuesday, March 6, 2012, Allen S. Rout
A month or so ago, I posted about some slow focus updates I was seeing in
xmonad.
http://article.gmane.org/gmane.comp.lang.haskell.xmonad/12074
I noticed another correlation when this is happening, and thought I'd jot
it down in this forum, in the hopes that it eventually contributes to a solution.
During the period when the focus is slowly traversing a lot of
intervening windows, I've found that I peg a CPU running the X server; My Xorg process goes to 99%CPU and stays there for the entirety of the delay.
I've been watching the X cpu behavior because I've also noticed that some
Firefox browser behavior will similarly peg the Xorg process for seconds at a time, and I was trying to suss that out.
I'm kind of free-associating here, but I'm wondering if xmonad's use of X
resources is leaving Xorg with a long list of [wumpus] that the X server can't collect or collapse or something, I have no idea what the wumpus might be, but...
- Allen S. Rout - Climbing back under his rock
_______________________________________________ xmonad mailing list xmonad@haskell.org http://www.haskell.org/mailman/listinfo/xmonad
On 03/06/2012 09:31 AM, Don Stewart wrote:
Does it happen with an out of the box xmonad ? With no user config?
Sorry to take so long to respond: I don't close X sessions lightly. But my home laptop's power brick futzed on me, so a reboot happened, like it or not. :) The short answer is "Yes". Long answer follows: I made another user and logged in with base xmonad as supplied by stock oneric ubunutu, made 20-mumble XTERMs on one screen and a top in the primary window. The machine was snappier to respond than my workstation at work, but the laptop is simply newer, too. (plus it was a brand new login, nothing else going on on the machine, etc. etc. etc.) But I could easily push the Xorg CPU to the 90% and above by simply moving my mouse across the windows, and when I did it vigorously, I could get seconds ahead of the focus updates. (i.e. stop moving the mouse and watch focus rattle between the XTERMs over and over again while it caught up). For giggles, I then started base stock CTWM, which was my favored platform before. I started ~40 one-character-wide XTERMs and lined them up next to each other and wiggled my mouse over them. Strictly speaking, there were probably something like twice the number of transitions there, because each xterm->xterm transition was really an xterm->root_window->xterm transition. Or maybe even xterm->border->root->border->xterm. I dunno. Anyway, wiggling my mouse over that field, I couldn't get my Xorg CPU above about 10%. - Allen S. Rout
participants (2)
-
Allen S. Rout -
Don Stewart