Idle OmniWeb consuming 20% of CPU cycles
 

OmniWeb is currently idling with a single, blank page open in my default workspace (no pages saved).

I've been checking Activity Monitor for the past 10 minutes -- OmniWeb is consuming a solid 18-21% of the CPU ... doing nothing.

[Several minutes later: When I quit and relaunched OmniWeb, the CPU load dropped to 0.00%.]

[SIZE="1"][copy of my formal bug report][/SIZE]

Next time this happens--would you please try typing the following command in a terminal window:
[CODE]sample OmniWeb 2[/CODE]
it will then capture what OmniWeb is doing every 10ms for the next 2 seconds and write a file
[CODE]Sampling process 10922 each 10 msecs 200 times
Sample analysis of process 10922 written to file /tmp/OmniWeb_10922.sample.txt[/CODE]
Please then send feedback (Help->Send Feedback) and include the file.

To include the file in mail quickly, click the attach button and when the open sheet comes up, type a '/'. The "Go To Folder" dialog will pop up and you can finish typing "tmp" and hit return to go straight there.

Let me know how you do this, Ward. It's a great tip. I can't seem to get mine below 70% - 20% would be killer.

;)

I have experienced this also. There was one time when my PB's fan would not go off. Activity Monitor reported that OW was using 20% of CPU, even though I had no pages open. After restarting OW, it went back to 0.

I will try to report it next time I notice it.

nasty load
 

After years of this bothering me I have decided it's likely a javascript problem. Specifically: if I visit a page with lots of js perhaps it loads something into OW which can't unload it when the page is no longer around. Thus restarting, which doesn't load the offending page, is a "cold boot" GC.

It could be some other plug-in (flash sucks pretty hard too). The reason I think it's JS is that I mostly have JS disabled except for certain pages. It seems to happen when I have visited those pages.

But I don't know the internal architecture of OW so this is a black-box analysis. So don't treat it as any kind of massively-researched gospel!!

From what I have seen in other threads from OmniGroup personnel, this is linked to Javascript and/or garbage collection (whatever that is). I'm hoping that they get it fixed before 5.5 final (preferably for the next beta would be nice!) as this issue has been really annoying to me ever since the first sneakypeek.

[QUOTE]I'm hoping that they get it fixed before 5.5 final (preferably for the next beta would be nice!) as this issue has been really annoying to me ever since the first sneakypeek.[/QUOTE]

Honestly, I'm not holding my breath. Every browser I've ever used has suffered from serious memory leaks and concomitant eventual slowdowns. OG identified a major bug related to JS, but it's just one bug on top of what I'm guessing are many, many more. Barring radical changes in development techniques and tools, priorities, the OS, or some combination of these, I don't expect this problem to go away.

That said, there is hope. A while back, the Safari team went on a crusade against memory leaks. They were actively soliciting feedback in this area, and they crushed literally thousands of memory-related bugs. That's wonderful, but I doubt they can keep it up in the face of other issues that are more visible to the average user, plus the demands of adding new features.

Something is running
 

While memory leaks are a chronic problem with many applications, I doubt it's the CPU hog when nothing is happening (e.g., blank window, no window).

Something seems to be running. gumby's suspicion about JavaScript makes sense from the black box vantage point.

I hope I don't have to be reminded too many more times to sample a run-away application.

CPU hogging is not a memory leak (unless the leaking is a process that sucks CPU cycles). I would say that it has to be a priority for OmniGroup to fix the majority of the CPU hogging before final - people are not going to appreciate having to quit the browser every hour or so just to return their computer to a usable state.

I never ever got this problem with OmniWeb. In no version. Since years. But always with Safari and all the Mozilla stuff around.