Sayrus

I've seen VISA cards with several banks in France where there is commission after 1 to 3 monthly ATM so I'd be doubtful about VISA having such as requirement.

At least in France, most of what people call "credit cards" are actually debit cards.

Historically, AWS own infrastructure relies on us-east-1. Loosing us-east-1 usually means loosing many other AWS Global services which are required for services in other regions to be healthy.

> Tokens saved are tokens saved.

Not always. RTK strips flags and other information. Sometimes you spend more tokens getting them back later. Sure your saved 70% tokens on that tool call, but nothing in the metrics says whether you ran 3 tool calls instead of 1.

There is also a question of whether that stripped output requires more thinking tokens or not.

I'm not sure how Garmin works, but for instance with Google Wallet-compatible watches, you need a phone where wallet can run. I've had this setup for a year where I loaded the cards from another phone and used a watch to pay.

However Wallet didn't like this setup. Tokens expired at varying delays, sometimes a day, sometimes a week or payment failed without reasons.

Nowadays, I just use my bank's app which work fine on GOS.

Excluding server costs, having that 100Gbps on egress can cost $50k a day. since it's a very high-margin product, AWS support would probably refund or reduce that to hundreds. Not sure how you get to millions either.

The data-sharing surely is for all providers. I think the sentence "When models become available, onboarding details will be shared." hides a lot of things.

> Through Google Cloud's Agent Platform: Retention will need to be enabled for your new covered model, and retained data stays in your GCP environment. When models become available, onboarding details will be shared.

From https://support.claude.com/en/articles/15425996-data-retenti...

MV3 itself isn't what breaks uBlock Origin, it is the bundled removal of capabilities that Chrome decided to do. Firefox MV3 supports full WebRequest "scopes" while Chrome only supports declarativeNetRequest.

Most panels are from China. Panels have a very long lifetime. Over their lifetime they generate way more than their price in oil. Europe is not a huge producer of oil and relies on imports to sustain its usage. Sourcing panels is effectively reducing the amount of money leaving Europe in the long term.

> Now that the attack window has changed to 7 days, all new exploits like these will come with time bombs to not trigger until 8 days.

Many automated scanners use static code analysis rather than run the installation script. Not all of them are caught, but a good part of them are and you'd be saved by a delay.

You still need criteria to handle reputation: does an account invited years ago and now spamming affects the reputation of the inviter, how much? What about the hacked accounts?

For small platforms it makes a lot of sense, for larger the potential for abuse is still there in different forms.

> Gecko doesn't have a WebView implementation (GeckoView is not a WebView implementation), so it has to be used alongside the Chromium-based WebView rather than instead of Chromium, which means having the remote attack surface of two separate browser engines instead of only one. Firefox/Gecko also bypass or cripple a fair bit of the upstream and GrapheneOS hardening work for apps. Worst of all, Firefox does not have internal sandboxing on Android.

> The sandbox has been gradually improving on the desktop but it isn't happening for their Android browser yet.

Context is definitely interesting to have with your statement (From https://grapheneos.org/usage).

Thanks for writing all this, this really shows how the failures you encountered don't overlap with my use of the phone.

I don't use RCS and Android Auto.

I have HeliBoard to replace Stock/Google Keyboard. It is way ahead the stock keyboard experience but far behind Google Keyboard's, especially when writing in two languages.

Tap-to-pay works with my bank apps. But that means I can only use one card unlike with GPay.

I rarely use second account as the latency to switch from one account to the other is a pain. I only have a secondary sending notifications to the first one.

I don't let the phone auto-reboot for installs, I let it install automatically and click reboot when I want it to install.

I am on a physical SIM / different carrier and never encountered network issues so I can't comment on that one.

I think parent is talking about Play Integrity being integrated into banking apps. It's a hit or miss depending on the bank, some will be fine without, some with integrate it but not rely on it to directly refuse login, some will require a lower integrity level, and some will actually require the highest integrity level leading to issues on custom ROMs.

I've been using it for a bit over a year. Installed in a few minutes thanks to WebUSB. A bit of research needed to set the right permissions on Google Play Services.

After that? I only had one application fail due to Graphene's memory allocator. No weird bugs, no need to restart like some siblings are commenting. As close to the "Graphene just works" as it could be.

However, I'm not heavy into Google's ecosystem. Google Pay will not work but I'm not a user, some Google features won't tell you why they don't work but I'm not using them either (Quick Share for instance), none of my apps require the highest Play Integrity level. Maybe the person who say this are a specific type of person where use-cases don't overlap with what breaks on Graphene.

Not sure what country you're from but in France it's not rare to see 0.30-0.60€ per kWh and even requiring a subscription on top of that.

ARR says nothing about the ability of these companies to retain customers once subsidies stop.

An unfinished upload or sync stored locally instead of on the remote can absolutely be an issue. You can look at all the posts about OneDrive and GDrive not actually syncing before confirming to users who then delete their files since they "have been uploaded". Or the user may never open that specific page again or the session will not exist anymore when he comes back.

Browser storage is cheap, but it is not guaranteed to be durable.

On the other hand, "are you sure you want to exit without saving" is a good use-case. But I'd prefer that to be a setting I can allow for specific site.

$99 per year for your developer account required to distribute applications. At AWS pricing, that's a bit over a TB of traffic. At any normal pricing, that's anywhere from 10TB to a few hundreds. At volume pricing, that's even more. How many apps are paying for traffic they don't use? Apple pockets millions.

> If you are analyzing locally stored transcripts, you wouldn't see raw thinking stored when this header is set, which is likely influencing the analysis. When Claude sees lack of thinking in transcripts for this analysis, it may not realize that the thinking is still there, and is simply not user-facing.

Claude often fetches past transcript for information after compaction. Wouldn't this effectively distort the view it has of past discussions?

And your car, license and insurance are not such a connection?

Extrapolating with this method, large cities should be as well equipped as Switzerland.

Los Angeles has 10x the population density of Switzerland and around the same population density in urban area. Same for Washington dc, San Francisco...

It doesn't have to be an all-or-nothing. Distance between cities may be large, but the order of magnitude is closer when comparing within cities.

I think what they did with first-party support for LSP would be an example of this.

However, Neovim explicitely states that they don't want to turn VIM into an IDE. The feature parent is talking about seem to be falling into that type of vertical integration instead of composability.

Emergency situations are defined by two situations: severe accidents and manual press of the button. Article 6 covers the data being sent, Article 5 covers the manufacturer's obligations. Your audio during the accident and your last three locations may leak, but the eCall system is not designed for a permanent phone-home system. If I remember correctly, you can't even use the eCall SIM for tracking as that'd encourage people to disable safety features.

All the things you are talking about, permanent phone-home, tracking of location, audio and video, driving habits, are tracked, sent and resold. That's what smart cars do. But it is not done through the eCall system. See it from the company perspective: why would they risk penalties for non-compliance when they can gather and resell all personal data with no risks using their own system instead of a safety one?

Happy to read your thoughts, can you elaborate on this?

Source for the amendments discussed:

- https://www.europarl.europa.eu/doceo/document/TA-10-2026-007...

Which is exactly my point. Cars are reporting on you, but tying that to remote assistance is disingenuous.

I keep hearing this one. But at least for EU, the eCall system requires external communication to be disabled until activated during serious accident. It cannot be used for tracking the vehicle in real-time.

Some parts of the legislation (https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:32...):

> 2. The personal data processed pursuant to this Regulation shall only be used for the purpose of handling the emergency situations referred to in the first subparagraph of Article 5(2).

> Manufacturers shall provide clear and comprehensive information in the owner's manual about the processing of data carried out through the 112-based eCall in-vehicle system. That information shall consist of:

> the fact that there is no constant tracking of the vehicle;

That vehicle nowadays are equipped with always-on internet and microphones is not related to remote assistance.