That'd be fine, but why did Firefox (Mobile) go ahead and cache it? When it hadn't cached it all those other times it had downloaded fully?
Announcement
Collapse
No announcement yet.
Mozilla's WebRender Making Good Progress, Can Be Tested On Firefox Nightly
Collapse
X
-
Originally posted by Michael_S View PostInteresting. I run Firefox and Chrome with EFF Privacy Badger against the benchmarks on browserbench.org every few months, and Firefox wins big in one, loses by a tiny margin in two, and loses big in one.
Privacy Badger is not an adblocker, but a privacy guard that works only on third-party tracking. It locks only third party stuff it finds is actually tracking you. Or so it says on its webpage.
I think this means Privacy Badger has less work to do than uBlock origin, so it's normal that it has less impact on performance.
Comment
-
Originally posted by PluMGMK View PostThat'd be fine, but why did Firefox (Mobile) go ahead and cache it? When it hadn't cached it all those other times it had downloaded fully?
Mobile is all about conserving power and 3G/4G modem connection is power-intensive, and also must take into account shitty connection speeds due to bad signal reception (you are in a building, in a fast-ish moving vehicle, and so on), overcrowding the same "cell" (the antennas serving 3G/4G in an area).
So they cache like crazy, because cache (either RAM or flash) is faster and less power-intensive than waking up the modem from sleep to send/receive stuff.
As others said, the failure to update the cache is actually a server/site issue. Decent websites provide versioned assets, so the browser can know beforehand if it needs to pull stuff out of cache or download again.
EDIT: also the overwhelming majority of mobile connections have a very low GB quota, that's another big reason to cache like crazy.Last edited by starshipeleven; 27 November 2017, 05:53 PM.
Comment
-
-
Doing this right is going to require attention to privacy and fingerprinting issues. The use of GPU rendering in the canvas drawing API is a notorious method of cookieless tracking, forcing many users both to use a canvas blocker/randomizer and to turn off GPU acceleration. To make this not trackable would require that the server be able to render content with GPU acceleration, but not be able to send any of that content nor a hash of it (etc) back to the server so that the slightly different results of different GPUs and drivers could not be tracked and logged.
Thus it is my hope that Mozilla will ensure that Webrender cannot be used to create anything that can be returned back to the website server, and to make this the default for all users so blocking a return of information is not itself uncommon and thus fingerprintable. When the US government's prosecutors say (in the Inauguration protest cases) that visiting a website can be grounds for conspiracy charges, this is really fsck'ing important!
- Likes 2
Comment
-
Originally posted by PluMGMK View PostThat'd be fine, but why did Firefox (Mobile) go ahead and cache it? When it hadn't cached it all those other times it had downloaded fully?
Comment
-
Originally posted by tessio View PostSo we'll finally get GPU rendering by default on Linux?
Webrenderer just goes extra steps making almost everything being drawn on GPU (I think no browser have figured out how to do fonts without rasterizing just yet).
- Likes 1
Comment
-
Originally posted by swoorup View PostWaiting on the next generation of desktop apps using web technologies powered by Webrender
Comment
-
Originally posted by tpruzina View PostYou are already getting that to varying degree, bunch of stuff is already accelerated unless your card/driver is blacklisted due to bugs (which won't change).
Webrenderer just goes extra steps making almost everything being drawn on GPU (I think no browser have figured out how to do fonts without rasterizing just yet).
[Tracking bug] [Linux] Turn on OpenGL accelerated layers by default for at least some subset of hardware
Comment
Comment