[Android] Viewport Size protection makes site loading very little long

[HarriBuh]

Since the introduction of the Viewport Size protection almost every page one the www (almost = not sure whether it is always the case or not) takes a minute to fully load.
When I turn the option off, everything is fast and well again.
Is this the sacrifice for improved security or just a flaw of the feature that will probably get fixed anytime soon?

[uazo]

Since the introduction of the Viewport Size protection almost every page one the www (almost = not sure whether it is always the case or not) takes a minute to fully load.

could you try this version?
#50 (comment)

Is this the sacrifice for improved security or just a flaw of the feature that will probably get fixed anytime soo

if it is a problem with the feature, I will try to fix it. if you can, try that version first, in case it doesn't solve it, let's better understand where the problem might be.

[HarriBuh]

Yes, my version is the/ your latest.

[uazo]

OK, so some more information is needed, both about the device and the sites where you detect the anomaly.

[HarriBuh]

Google Pixel 7
Android 13
Rooted

Randomly visiting pages on the wwww:

www.brewcompany.de (took very long to load)
www.gsmarena.com (took very long to load)
www.whathifi.com (loaded some images fast, but took long to finish)
www.urbandictionary.com

All of them took some time to load, but I wasn't able to fully reproduce my issues.
It almost seems that the longer I keep the browser running, the better it gets.

And it seems like pages with a lot of content/ images are much more often affected than text-only pages.

[Arien02]

Almost instant loading here with last test version.

[uazo]

Almost instant loading here with last test version.

Is the same thing that I also detect.
theoretically, viewport protection does not act on the network stack, whereas the bug I introduced with #51 does.

@HarriBuh confirm that you have tried https://github.com/uazo/bromite-buildtools/releases/tag/TESTONLY ?
I should be able to compile #66 tonight, maybe we can check tomorrow.

[HarriBuh]

@uazo I can confirm that the issues I had are gone since using your TESTONLY version. You fixed it 🔨👍

[uazo]

perfect!
a question addressed to you all: the path modifies the viewport by a few pixels, so that it is detected by javascript as different at each first party.
I developed it in such a way that the user does not notice the difference in page rendering, do you detect any difference?

[HarriBuh]

I had an issue while loading a image-heavy page once or twice. The page would not load the correct order of the images and text boxes (sorry for my bad vocabulary, but I'm no IT guy), much like it would do when blocking Javascript completely. Several manual reloading triggers solved it.