There is a new version on Github
uBol-stripped/uBlock-Stripped-Dynamic-7_8_1.zip at main · Kees1958/uBol-stripped
I have tested it and it seems to work okay, this is a big refactor
- background.js restructured from uBlock-lite to uBol-stripped programming standards (about 1/8 code reduction and better maintainable)
- big chunk (AG) vs small chunk (uBol) approach
a) DNR-rules I am using big chunk for DNR rules, since Chrome's internal adblocker handles them crazily efficient, I will stick to AG approach
b) Scriptlets - still using the big chunk approach (but in terms of page injection size we are only talking about 600kb, so it is minor) but only using AG scriptlets internally and using a different allocation mechanism reduces V8 precompiler overhead.
c) Cosmetic filtering - now using the small chunk approach for the javascript injection overhead needed for cosmeting filtering to work (so not for the actual cosmetic rules themselves, to prevent FOUC)
These are minor changes in terms of page injection size (reduces from 2.2 MB to 2.1 MB), but major code changes, because it allows the V8 javascript interpreter to work more efficient. For reference when I started to add AG skills to uBo-lite the page injection size was 6.8 MB. So I have been able to put the uBO-lite efficiency (R. Hill's brilliant solutions) while adhering AG team professional programming standards.
TLDR:
1. a lot of changes under the hood
@Jan Willy could you try this version V7.8.1 (because you are using uBlock stripped in a more advanced way than I do
)?
2. because something changed from sync to asycn I want to make sure I still have the AG advantage of no FOUC, so
@Sampei.Nihira could you give 7.8.1 a try)?
These changes result in V8 to compile 90% less code per page load. That sounds spectacular but it has marginal effect on the Speedometer 3.1 benchmark.To be honest that was a disappointment for me and probably the reason the AG team choose their approach (it is a lot easier to have access to all data all the time, allows for simpler programming structure).
It is like those synthetic benchmarks when you buy a new PC on paper it is hundred or thousend times faster than your old PC, but in practice it only is a second faster in page loads. Only when you do intensive work (like graphics design or gaming), you feel it is (a lot) faster.
