µBlock, a lean and fast blocker

Discussion in 'other software & services' started by gorhill, Jun 23, 2014.

  1. elapsed

    elapsed Registered Member

    Joined:
    Apr 5, 2004
    Posts:
    7,076
    You aren't proving anything other than your willing to lie to your users, I wonder what the hidden agenda truly is with you.

    I was sceptical about this addon originally after you went out incorrectly criticizing EasyList's whitelist in your HTTPSB thread a while back, followed by me proving you wrong. Now I'm even more sceptical after your attitude towards this. (I already brought up how measuring blocking power itself is a flawed idea in the previous thread, yet you continue trying to push it like you have something to gain).

    At least the ABP devs don't openly lie about the competition in order to get installs.
     
  2. gorhill

    gorhill Guest

    Yes, I remember. I did feel bad about having presumed wrong, and still feel bad about it.

    You brought me your opinion that you think it the benchmark is flawed. I just disagree that it is flawed.

    Try this thought experiment: imagine I hide the specific lists, and just coallesce all of the out-of-the-box lists used into four categories, "Ads", "Malware", "Trackers", "Social". Then for the benchmark I would check "Ads", "Malware" and "Trackers", just like I did for the others blockers. Results of the benchmark would be exactly the same as what I obtained, and which results you don't like.
     
  3. gorhill

    gorhill Guest

    I've given more thoughts on this. Since hits on malware lists are likely not happening, the only list left to explain the difference between uBlock and ABP is Peter Lowes' Ad Server list. I've been supporting this one out of the box since the beginning in HTTPSB, it was the first 3rd-party list I supported if I recall.

    I went back to Peter Lowe' site today, and I see that it is possible to get a URL to that list in ABP 1.1 format. Thus I was able to add the list to ABP. So I wondered whether I should include it in the next benchmark or not. I decided I won't, because of the not so trivial extra steps involved.

    One of the developer of Disconnect wasn't also happy with the results of my benchmarks, and in his case he thought the flaw was that I should not have changed anything after installing the extensions (specifically Ghostery), which I thought was silly as it would prevent making any meaningful benchmarks at all.

    So there is a middle ground between not fine tuning anything at install, or having to go through no so trivial extra steps to reach a specific configuration. I choose the middle ground.
     
  4. demoneye

    demoneye Registered Member

    Joined:
    Dec 30, 2007
    Posts:
    1,356
    Location:
    ISRHell
    this just remind me this ,there are lots of list tagged after Ublock is install by default (7 malware domain at the top caught my attention ).... is all this malware domain blocking is a must to be tagged ? i KNOW to many list ticked lead to slow browsing.... what your take on that ?


    cheers :)
     
  5. tlu

    tlu Guest

    Why don't you continue to use HTTPSB as a mere adblocker (go to the "About" tab and click "Start from scratch")? You can easily add individual exception rules.
     
  6. gorhill

    gorhill Guest

    Depends of the implementation. I am confident that it's not an issue with uBlock. I still want to comparatively measure the overhead for each net requests for popular blockers, but what I found so far with HTTPSB (which has a more complicated filter evaluation codebase) is that the overhead is significantly lower than ABP, and given how ABP filters are implemented in HTTPSB/uBlock, adding filters does not proportionally increase overhead -- I expect the increase to be marginal. But I need to measure just to get objective numbers, and that part is a bit time consuming as I have to add benchmarking code to other extensions.

    Edit: I am working on something unrelated, and I wanted to know the time spent in the filtering engine on average, and after having enabled Fanboy's lists (two of them) + hpHosts (almost 20,000 filters), I measure that on average it takes less than 0.2ms to evaluate a URL against the net filters (66,842 of them). So far, lots of filters is not an issue.
     
    Last edited by a moderator: Jun 27, 2014
  7. JohnMult

    JohnMult Registered Member

    Joined:
    Mar 26, 2012
    Posts:
    133
    Location:
    Greece
    After the last update I noticed without changing the filter lists I had chosen that in some sites for example www.contemporist.com appears a grey square in the place of ad that before was not. Same thing is happening and in other sites.
    BTW your adblocker is exactly what I wanted for Chrome. Sincerely great work!
     
  8. gorhill

    gorhill Guest

    Yes, sorry about that, I needed to reverse the solution I had come up with to blank blocked elements. I forgot to add this important note this in the changelog. I think collapsing blocked element is something I will work on before addressing other issues.
     
  9. Jarmo P

    Jarmo P Registered Member

    Joined:
    Aug 27, 2005
    Posts:
    1,207
    I do like the name ublock. Can't bring myself to remember ways how to produce the greek alphabet that is used for friction mostly in science. But yes this sounds nice, whether pronounced with or without. I guess running HTTP Switchboard I would not need it?

    I might install it though. Sometimes HTTPSB surfing is a bit restrictive and instead to turn it off, maybe your new thing as careful as I am to not install any extensions in Chrome.

    Keep up the good work Ray.
     
  10. demoneye

    demoneye Registered Member

    Joined:
    Dec 30, 2007
    Posts:
    1,356
    Location:
    ISRHell

    Ok i understand it this way or another , i just want to add this the filter isuue ... filter names "Fanboy’s Enhanced Tracking List" mess up some sites when enabled , i dont know if it related to Ublock or to this filter only , coz i cant locate this filter in fanboy site (maybe its only me) .

    cant u check this out?
     
  11. gorhill

    gorhill Guest

    I need more specific details. I need a specific site, and what part is broken, so I know where to start to investigate.

    The Fanboy's Enhanced Tracking List there: http://www.fanboy.co.nz/filters.html
     
  12. demoneye

    demoneye Registered Member

    Joined:
    Dec 30, 2007
    Posts:
    1,356
    Location:
    ISRHell
    according to fanboy site regarding this list -> Enhanced Trackers List (could cause some false positives, please report any issues)

    so i assume like i said its not your Ublock fault... and i dont want to bother you with site who got messed up :)

    10x man :thumb:
     
  13. J_L

    J_L Registered Member

    Joined:
    Nov 6, 2009
    Posts:
    8,738
    The more streamlined, the better. Also noticed HTTPSB and I presume uBlock doesn't allow you to add custom lists (Anti-Adblock Killer), which is why I'm keeping ABP for now.
     
  14. demoneye

    demoneye Registered Member

    Joined:
    Dec 30, 2007
    Posts:
    1,356
    Location:
    ISRHell

    GOOD POINT....


    it be awesome if Ublock support anti adblock killer ... some site over here block me :)
     
  15. Raymond,

    Adding malware URL-filtering does not add much real protection, when I use Chrome's build-in malware protection. Also when testing blocking efficency (based on reported blocks by uBlock itself) running some user simulations with cache clearing and re-boot between sessions, this seems to be the most effective setting (memory/delays/blocking/false positives). When national list are available these could be used.

    Average user is not a paranoid user and 9 out of 10 leave the "phising and malware protection" at default anyway (enabled). Efficiency (fast & light) and effectivesness (NO acceptable ads whitelist and extra blocklist in default mode), should provide enough arguments to choose your adblocker above others.

    Could you run some tests yourself with your own test tool using these "default" settings and compare it again to others in default settings.

    GUI enhancement: could you put in some header tags to make the list easier to read?

    Default block list:
    -
    -
    -

    Extra malware blocklists
    -
    -
    -

    Extra tracking blocklists
    -
    -
    -

    Extra adds blocklists
    -
    -
    -

    Additional country blocklists:
    - Dutch
    - German
    - Etc

    Regards Kees
     

    Attached Files:

    Last edited by a moderator: Jun 29, 2014
  16. JohnMult

    JohnMult Registered Member

    Joined:
    Mar 26, 2012
    Posts:
    133
    Location:
    Greece
    After the last update everything is back to normal. Thanks a lot for sharing with us your blocker. I think its the best of its kind in Chrome.
    One question only in site for example www.videocosmos.gr I have add the filter gr##.sidebar-dia in my filters but its not working and I noticed that although I have 4 filters in the 3rd party filters is counting only 3 apparently this one. I think the syntax of the filter is ok. Am I doing something wrong?
     
  17. gorhill

    gorhill Guest

    I like the idea of headers for grouping lists: https://github.com/gorhill/uBlock/issues/30

    For tests, sorry but these are very time consuming, with always a lot of details to check beforehand to be sure the benchmarks are not going to be "tainted". The tool I used for benchmarking is available at https://github.com/gorhill/sessbench, and the new reference benchmark script I used in this tool is at https://github.com/gorhill/uBlock/wiki/Reference-benchmark, so that anybody can reproduce the ones I make, or create their own.
     
    Last edited by a moderator: Jun 29, 2014
  18. gorhill

    gorhill Guest

    Excerpt from ABP filter doc:

    uBlock implementation also requires full domain name. So "gr" is not an effective domain name, you need to provide the whole domain name on which the filter needs to be applied, that is, "videocosmos.gr" in your case.
     
  19. Good idea to make it a repeatable test users can test themselves, stops endless discussions and theorizing, everyone can test for themselves now
     
  20. Rasheed187

    Rasheed187 Registered Member

    Joined:
    Jul 10, 2004
    Posts:
    17,546
    Location:
    The Netherlands
  21. gorhill

    gorhill Guest

    No clue, I never looked into Opera 12's API. Somebody would have to look into this, I am currently way to loaded, with no end in sight.
     
  22. Rasheed187

    Rasheed187 Registered Member

    Joined:
    Jul 10, 2004
    Posts:
    17,546
    Location:
    The Netherlands
    OK thanks, I can understand that you´re quite busy. ;)
     
  23. gorhill

    gorhill Guest

  24. Sordid

    Sordid Registered Member

    Joined:
    Oct 25, 2011
    Posts:
    235
    yeah, muBlock is faster than adblock/plus in my tests. Nice work, Gorhill. I will still use HTTPSB though--much more robust, albeit slower.

    Does anyone have a method to benchmark browser extension cpu time other than just generic run/avg or viewing the task manager/s?
     
  25. dogbite

    dogbite Registered Member

    Joined:
    Dec 13, 2012
    Posts:
    1,290
    Location:
    EU
    Installed right now. I want to see how behaves compared with AdGuard.
    Thanks gorhill for this new work.
     
  1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.