View Poll Results: After a few days worth of testing, is this patch helping you?

Voters
7. You may not vote on this poll
  • Yes, I'm not having to restart as often and get the too many cards error less often

    3 42.86%
  • No, I'm still frequently getting too many cards even with this patch

    1 14.29%
  • I wasn't getting the too many cards error to start with!

    3 42.86%
Results 1 to 13 of 13
  1. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #1  
    This patch requires additional testing - but it seems to restore the old memory garbage collection method that was in webos versions before 1.3.1. In my testing (2 days worth), it fixed the memory leaks associated with using drPodder & Newsweek.

    Please test - especially if you're frequently getting the too many cards error after updating to webos 1.3.1.

    What it changes is

    Code:
    # Automatically remove eventlisteners from DOM nodes not attached to
    # the render tree.
    AutomaticallyCleanupEventListeners=false
    to

    Code:
    # Automatically remove eventlisteners from DOM nodes not attached to
    # the render tree.
    AutomaticallyCleanupEventListeners=true #default is false
    My observation on what that means is that garbage collection is cleaning up after apps instead of relying on the apps themselves to do it. But again, please test.. This is just based on 2 days of testing from me. (I was getting the too many cards error at least once or twice a day.. been running great for the past 2 days using this patch) (this might re-introduce the 300ms lag in games that some were complaining about in webos 1.2.1 - I never saw that lag so can't comment on that..)

    -mark

    PS - the patch has been submitted to preware.
    Attached Files Attached Files
    Last edited by diomark; 11/22/2009 at 01:55 PM.
  2. atom00's Avatar
    Posts
    72 Posts
    Global Posts
    75 Global Posts
    #2  
    Looks very interesting. One question: Can this patch be used alongside the lower swap threshold and extra cards patches?

    Thanks again for all the great work.
  3. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #3  
    Quote Originally Posted by atom00 View Post
    Looks very interesting. One question: Can this patch be used alongside the lower swap threshold and extra cards patches?

    Thanks again for all the great work.
    They're independent from each other.
    -m
  4. Quintus's Avatar
    Posts
    624 Posts
    Global Posts
    672 Global Posts
    #4  
    I'm really looking forward to testing this out as the my experience with 1.3.1 has been frustrating. At this time the patch is not available on Preware.
  5. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #5  
    Quote Originally Posted by Quintus View Post
    I'm really looking forward to testing this out as the my experience with 1.3.1 has been frustrating. At this time the patch is not available on Preware.
    It's been submitted, but it takes time for them to go through requests. You can install it via WebOS QI though using the patch file in the first post. (just drag it over to the quickinstall window and hit install)
    -mark
  6. xorg's Avatar
    Posts
    633 Posts
    Global Posts
    1,010 Global Posts
    #6  
    I noticed this parameter is in the browser.conf file, which is likely loaded by the BrowserServer service. The too many cards issue is likely related to LunaSysMgr retaining too much resident memory even though cards are closed (memory leak likely). So I'd be surprised the parameter changes things unless it is related to LunaSysMgr. Could be.
    xorg tools - power tools for power users
  7. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #7  
    Quote Originally Posted by xorg View Post
    I noticed this parameter is in the browser.conf file, which is likely loaded by the BrowserServer service. The too many cards issue is likely related to LunaSysMgr retaining too much resident memory even though cards are closed (memory leak likely). So I'd be surprised the parameter changes things unless it is related to LunaSysMgr. Could be.
    Actually, the browser settings are controlled by /etc/palm/browser-app.conf

    The browser used for webos apps (i.e. any other app) is /etc/palm/browser.conf

    (don't forget WebOS apps are basically html/javascript/css apps - which run on the webos internal 'browser' on the device - vs. the actual web browser which seems to be controlled via /etc/palm/browser-app.conf)

    If you run strings on LunaSysMgr, you can see that it reads the browser.conf file -

    Code:
    $strings LunaSysMgr  | grep '.conf'
    /etc/palm/browser.conf
    /etc/palm/luna.conf
    /etc/palm/luna-platform.conf
    user-confirmed
    /etc/palm/notificationPolicy.conf
    /etc/palm/persistentWindows.conf
    /etc/palm/lunaAnimations.conf
    /etc/palm/lunaAnimations-platform.conf
    $
    (but the BrowserServer looks in browser-app.conf)
    Code:
    $strings BrowserServer  | grep '.conf'
    /etc/palm/browser-app.conf
    /etc/palm/browser-app-platform.conf
    $
    -mark
    Last edited by diomark; 11/22/2009 at 01:44 PM.
  8. #8  
    where is the link to download this?
  9. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #9  
    Quote Originally Posted by patricksmangan View Post
    where is the link to download this?
    Sorry.. Attached now to the first post.
    -m
  10. #10  
    It seems like a lot of the patches being developed are simply uncommenting lines of existing javascript code in the WebOS files, or making very slight changes (e.g., setting a variable from TRUE to FALSE).

    I wonder why so much of this stuff was commented out by the Palm developers, when turning it on seems to help a lot of users?
    cohoman

    Palm Apps:

    jVault and jChecklist
  11. diomark's Avatar
    Posts
    752 Posts
    Global Posts
    770 Global Posts
       #11  
    Quote Originally Posted by cohoman View Post
    It seems like a lot of the patches being developed are simply uncommenting lines of existing javascript code in the WebOS files, or making very slight changes (e.g., setting a variable from TRUE to FALSE).

    I wonder why so much of this stuff was commented out by the Palm developers, when turning it on seems to help a lot of users?
    Frequently in software development, developers will build something but QA/Testing will not have time to test that feature. My guess is that's why most of the options are there.
    -mark
  12. xorg's Avatar
    Posts
    633 Posts
    Global Posts
    1,010 Global Posts
    #12  
    Got it. Apologies for the misdirection. When things like that are shutoff, it could be because it's not ready to release yet. Hopefully it doesn't cause more issues than it might solve. Will be trying it myself though.
    xorg tools - power tools for power users
  13. #13  
    My memory might be going, but I think I remember seeing this set to False on all the previous webOS's. I thought it might be like that because of headless apps that still sit around catching events. Do you have a copy of webOS Doctor from 1.2.1 that you could poke around (I just realized I deleted all mine)?

    What I did notice, however, was that they removed the --gc_global from the [JavaScript] Flags in both browser.conf and browser-app.conf. From this thread, it seems that flag was responsible for the stutters.
    Quote Originally Posted by Brain_ReCall
    I'm an Embedded Software Engineer. My idea of a Good User Interface is printf().

Posting Permissions