changelog: v91.1 #1250
Comments
IMPROVEMENTSThanks for flying Delta. Hopefully we've bent the curve.
Thanks for pointing out typos and fuckups and the thumbsup in comments |
ALL HAIL PANTSCollect the set
|
|
this one took me quit some time to merge.(manually with Meld) |
|
So .. no comments/feedback (or emojis) on the improvements? How's everyone liking them? Does it make things easier? Do you like item Some feedback would be welcome. Don't make me sic the trash pandas on you all |
|
This will definitely make things easier. For people that don't bother reading and just take the user.js template as-is, this will reduce situations where people breakage. For my two cents, I'd prefer the |
|
Thanks. I don't think anything I did removes any meaningful breakage: wasm isn't used much. I don't think asm broke things, https alt srv didn't break things. no one uses graphite ... I was more concerned with people who DO read the user.js |
|
6050 was useful to me as i could reset/remove no longer used items. |
Line 1315 in 524823f |
🍻🍺🍻🍺🍻🍺🍻🍺🍻 I was going to wait till I'd gotten thru the whole thing (I'm only to 1600 atm), because holy &$#*@ this is a comprehensive overhaul. I see now why it took a minute to get released. Anyway, so far I'm not seeing anything that doesn't make sense. Only possible complaint is what @ginick said about lost explanations. Most of them I don't miss, but I was glad I got to read the stuff that used to be at the top of section 1000 and the notes on the old 1001 and 1023 before that info got axed. |
|
So .. no comments/feedback (or emojis) on the improvements? Onf |
But that's the point: when it was added it was relevant, today it is not
As for 1023, it only lost some info about "possibly" affected recently closed tabs/history - I can't vouch for that anymore. Its been years since I added that pref, and only tested it the once. This level of maintenance I can do without. The fact that cache is disabled saves magnitudes of order far more disk writes (if it has an impact for you) Now if there is something I removed that is still relevant, then we can put it back
they're still together (mostly: prefers-color and prefers-reduced-motion went to personal, and font pref was swapped out, whitelist vs vis) , just a different section number
Excluding section 7000s and 8000s (don't bothers) which I am not interested in, any links removed were a waste of time: e.g. the bugzilla which showed the pref being added, but it doesn't even have any info (seriously, some of them were only 4 comments long), and/or the pref/description is self-evident. I even replaced some, rather than remove them. Nothing of value was lost, I can assure you. And some replacements were better, e.g. the ASM and JIT ones e.g. I'll do some reference link number crunching and diffs |
|
I did some quick number crunching: excluding
diffs to follow v90: 228
v91: 149
|
here are your diffs 12 new references added in v91
91 references removed in v91
|
|
I'm not going to bother going through them - but the new ones replaced old broken ones, or were better (e.g. all the cve? ones). IMO nothing of value was lost. But there you go, now you have a list and can prove me wrong :) |
Awesome. Thanks for the explanation and the Two baby nits so far:
Fortunately things seem to be a little more humane around here :-) |
|
"Nothing of value was lost, I can assure you. And some replacements were better." 👍 |
|
I had the same concerns as the previous comments, but figured there were good reasons for the changes - thank you for explaining! I was able to remove a couple of custom overrides. Side-note, the override recipes are very useful - I feel like they should get more visibility somehow. |
|
A few more you might want to consider:
|
|
Sorry If I was harsh, I appreciate all you work!! But I cannot agree with all. Hail Pants, Earthlings and the others. |
equivalency: same result as something else: e.g. the fonts you use (or have available) on a webpage are dictated by the OS and by your preferred language - e.g. on Windows, Or more simply: let's just say most webaudio entropy stems from math entropy. You could randomize webaudio, but the entropy still remains in math. But fix math, you fix the webaudio - i.e webaudio is equivalency of math, and math is equivalency of OS, and you cannot hide your OS I don't actually understand the 8 year old argument made at Tor Browser - maybe it made sense back then (edit: they didn't have a font whitelist at the time?). If you block the web font, then you end up using your default font, which will likely need to fallback to another font (because generally speaking the chars in glyphs and icons will not be in your default font: e.g. PUAs), which then creates more entropy than if you had just let the web font load - which is why they also disabled the fallback rendering - so everyone would end up with tofu = breakage. Using a web font means all users use the same font and don't fallback anyway. So this never made any sense to me. Edit: by testing PUAs an attacker could probe system fonts which would vary mainly on linux (whitelisting didn't exist) Note: FF86+ 1676966: gfx.font_rendering.fallback.async
You can see this in tests, see https://github.com/arkenfox/TZP/issues/38 : until each Firefox session has come across a char it needs to fallback and resolve, it initially renders tofu, after which it remembers ... until the next session. Tests: (load each test in a new browser session)
TZP
|
Well the they refers to "cut, copy and paste". But it is a little awkward |
Hails are limited to PANTS
I didn't even understand your post
That's exactly what you're meant to do. The default user.js is a template and will not suit everyone .. indeed, anyone |
ESR users, get the v91.1 release
here is the v91-91.1 diff, which I added to the changes below
FF91 release notes
FF91 for developers
FF91 security advisories
CHANGELOG: [all changes ... 103 commits v90-91.1 ... good luck]
⭐ your friendly reminder to run prefsCleaner
ALL CHANGES (EXCEPT REMOVALS)
REMOVALS
STATS
The text was updated successfully, but these errors were encountered: