Replies: 5 comments 10 replies
-
Will there be a setting for this ? |
Beta Was this translation helpful? Give feedback.
-
Does that mean a reversal of this uAssets policy:
? |
Beta Was this translation helpful? Give feedback.
-
This should be announced in more places. Maybe at least some of these filters can be avoided. More filter authors should see this. Does anchoring to the start or end of an attribute helps at least a bit? |
Beta Was this translation helpful? Give feedback.
-
A few cms's use random chars either added at the start or end of the element, which is when There isn't many |
Beta Was this translation helpful? Give feedback.
-
When I open many new tabs, the hide rule uBlock Origin: 1.44.2
Chromium: 105
filterset (summary):
network: 98557
cosmetic: 128386
scriptlet: 23742
html: 0
listset (total-discarded, last updated):
removed:
plowe-0: null
urlhaus-1: null
added:
https://easylist-downloads.adblockplus.org/bitblock.txt: 31407-27, 2m
https://easylist-downloads.adblockplus.org/cntblock.txt: 1489-24, 2m
fanboy-cookiemonster: 40880-285, 2m
ublock-annoyances: 4569-52, 2m
default:
user-filters: 10-0, never
RUS-0: 42148-145, 2m
easylist: 64873-85, 2m
easyprivacy: 27492-533, 2m
ublock-abuse: 76-1, 2m
ublock-badware: 4222-99, 2m
ublock-filters: 32562-167, 2m
ublock-privacy: 251-8, 2m
ublock-quick-fixes: 421-2, 2m
ublock-unbreak: 1850-44, 2m
filterset (user): [array of 9 redacted]
modifiedUserSettings:
advancedUserEnabled: true
contextMenuEnabled: false
showIconBadge: false
modifiedHiddenSettings:
filterAuthorMode: true
popupFontSize: 13px
popupPanelHeightMode: 1
supportStats:
allReadyAfter: 753 ms (selfie)
maxAssetCacheWait: 690 ms |
Beta Was this translation helpful? Give feedback.
-
There are currently 1,137 highly generic cosmetic filters ("HGCF") with default filter lists, which are injected in every page/frame. There is no way to mitigate the impact these have by injecting them conditionally -- as was done with lowly generic cosmetic filters.
These HGCFs are a burden on the browser's rendering engine, for instance consider (picked four at random):
##[class*="-slot_ad-placements-"]
##[href*=".jetx.info/"]
##a[href*="//bongacams5.com/track?"]
##[src*="//www.dianomi.com/smartads.epl"]
You can see all those HGCFs for yourself using the More option in the Support pane, through the "CFE: Dump" button.
There is a lof of HGCFs which try to match using partial attribute value string comparison, and I suspect these are the worst offenders performance-wise on the browser's rendering engine. #1983 made the performance hit worse and more visible, but even with the reversion, I investigated further and found that specifically, the HGCFs are having measurable impact on even the most simple web pages.
What I have in mind:
By default, discard all HGCFs which are meant to partially match an attribute value. We now have a "Report a filter issue" button which make it easy for people to report content which should be filtered but is not. Given this, we would build our own filter solutions over time to minimize as much as possible that sort of HGCFs with performance under our control instead of leaving this to 3rd-party filter lists. Where a HGCF is really the only sensible solution, we can use
*##...
to make this explicit as these would not be discarded.Something else to investigate is that whether by dropping HGCFs by default we could make uBO enforce other generic cosmetic filters in Firefox for Android, it could very well be that HGCFs are the only one causing performance issue in Firefox for Android.
Beta Was this translation helpful? Give feedback.
All reactions