Hacker News Reader

Comment by lewisjoe | Hacker News Reader

Great news. One thing that I'm sure about V3 is that it isn't well thought-out at all. For example, imagine your extension has to cache data for a browser session (across tabs but with a single cache), it's impossible as of now.

The only workaround to do that has a 1MB storage limit, essentially forcing you to have a server-side cache mechanism (redis or whatever) for this trivial use-case.

And worse, Google developers essentially refuse to understand the problem - https://bugs.chromium.org/p/chromium/issues/detail?id=118522...

View on HN

This seems to be a pattern among Google employees, as least as I'm seeing from the outside.

They have their own company reality bubble, and clashes or inconsistencies with the outside world are met with disbelief.

The customer is always right - except when I think I'm smarter than their stated needs.

The customer, eg. the advertising industry.

Indeed, but it's a problem in engineering society-wide, not just Google. The more I think of it actually, it's just a human problem. Even kids default to this. I wish Socratic Ignorance were taught widely and often in school.

I remember getting into it with a Google person in a forum once. They kept insisting they couldn't do real, portable files in Google Drive for Docs because that would break collaboration. They didn't seem able to understand I didn't want collaboration. I wanted to be able to open my word processor/spreadsheet/etc files in other tools without needing to connect to the internet (predates offline mode). I wanted to be able to back them up somewhere (3-2-1 strategy). I wanted to be able to do automated analysis across all my writing. In short, I wanted to own my files.

They simply could not conceive of a use case that was offline or didn't trust Google to be reliable. I honestly don't think they ever used a real desktop office suite. They had no model in their head to understand it.

It is difficult to get a man to understand something, when his salary depends on his not understanding it.

That's exactly the saying that came to mind when I read the parent comment. These people have an agenda, the company one, to advocate for; and won't accept anything else.

Collaboration is the killer feature of the docs ecosystem. You will never convince them to break it for X reason.

I don't want them to break it. I understand why the two needs are incompatible. What I want is a way to opt out at a file or even account level so I can access my stuff in Drive. This is what they could not understand. It wasn't "this isn't possible," it was "why would someone want that?"

Export your Google doc to a word docx and then just save that in Google drive. It’s a portable file like any other at that point.

Nope. Not what I want. I want it to be able to work like office suites have since before Google existed. There is no reason it can't other than lack of priority on Google's part.

Microsoft's web-based stuff does this just fine, just like the offline stuff has for decades. I can edit on an app wherever I am, then have my laptop back home do some magic on the OneDrive folder and process/store/whatever I want, and I never have to invoke a magic menu item to ask Microsoft to let me. It's just there.

I don't see why I have to adjust my workflow for Google's failure of imagination. I've also had too many cloud things decide to limit/remove export functionality to trust it.

For this specific case of MV3, I wonder how many of those developers are actually writing and maintaining extensions. Yeah, I know one suspected motivation of this change is disabling ad blockers but IMO even if we ignore this aspect this migration is planned badly everywhere. If they really wanted to focus on just kicking ass of ad blockers, it would be done much quicker without facing this level of backlashes.

Not just for MV3 (or even Google), but many of those "API teams" actually don't have a good understanding on its actual use cases so the incentives are usually aligned with their own goals and directions rather than the actual customers because what they see everyday is just their code base and some OKR. I guess they really didn't want to miss out the opportunity to clean things up so they put every single wish list into the bucket without much user study and then it's spectacularly exploded as we know.

I've seen a bunch of "internal migrations", which is supposed to be a way easier than this kind of external ecosystem migration. Unless it was planned and executed very well, those teams are usually shocked by the initial backlash and how "creative" their users are. Sometime those teams are able to come up with reasonable compromises, but in many cases they just deny the reality then blow things up (which sometime works if there's not much dependencies though, but many case it's just wasted as soon as upper managements kick in). This is why almost all successful API migrations are accompanied with some sort of extensive user study from the beginning rather than some arbitrary metrics/goals set by themselves.

> For this specific case of MV3, I wonder how many of those developers are actually writing and maintaining extensions.

This is a huge problem, also for the Safari extensions team. Google doesn't ship many of its own extensions, and Apple ships none. Moreover, there's no evidence that either of these companies actually talk to extensions developers—i.e, have actual conversations with them—before shipping extension APIs.

Did Google talk with the developers of uBlock Origin or Tampermonkey before designing MV3? Not doing so is negligence.

Not just Google, all of Silicon Valley. Here on HN, I regularly see comments along the lines of: "Wait, do people actually still use Windows on servers!?"

Yes, yes they do.

In the last 10 years I haven’t had a single contract dealing with a windows server. Im sure that someone somewhere does indeed use it. But it’s gotta be 90%+ now a days Linux.

Selection bias. Do you work in Silicon Valley, at startups, or ex-FAANG places? Are you a "Linux person" looking for work related to Linux technologies such as NGINX, MySQL, Postgres, Apache, etc...?

Here in Australia it's a rare thing for me to see a Linux server. They exist, sure, but are generally in the minority.

I have some self-selection biases as well, but I regularly work for large orgs with more than 10K staff and I've been to orgs with over 1K IT staff, and in all cases Linux servers were 30% of the fleet or less.

By some[1] metrics, Windows Server is about 50% of the worldwide server OS market[2]: https://www.t4.ai/industry/server-operating-system-market-sh...

[1] Market research related to server OS usage is notoriously unreliable, with idiotic things such as counting "NGINX" load balancers as a "Linux web server" even if it is one tiny load-balancer VM in front a whole fleet of Windows IIS servers running ASP.NET. There are also the "hyperscalers" that skew distributions. The FAANGS alone add hundreds of millions of Linux instances, but in huge farms of essentially identical clones. At a single medium-sized customer, they have over 1K distinct Window Server VMs or HA pairs, but they also have 1x HPC cluster with 1K Linux nodes. Do you count that as "1" Linux install, or "1000"?

[2] Again, is it just counting the "for money" market, or just the "list of all systems used for servers, including the unsupported free distros?" Microsoft doesn't publish a separate line-item for their Windows Server revenue, for example, making it hard to determine how much it's still being used.

> Selection bias. Do you work in Silicon Valley, at startups, or ex-FAANG places? Are you a "Linux person" looking for work related to Linux technologies such as NGINX, MySQL, Postgres, Apache, etc...? Here in Australia it's a rare thing for me to see a Linux server. They exist, sure, but are generally in the minority

Huh, here in Norway, for big oil service companies, for government and electricity infrastructure, the wast majority of servers I see are Linux in the cloud.

Last time I saw a Windows server was a month before the pandemic, some old stuff that I inherited from another consultant that had been squatting there since last millennium I think.

Come to think of it, the only place I've seen a significant percentage of Linux was in an energy company in Western Australia...

I think you’re proving his point. That 10% (if true) is probably running a lot of important stuff in F500 companies. You just haven’t been exposed to it.

> In the last 10 years I haven’t had a single contract dealing with a windows server.

Windows is exceedingly rare, from a Silicon Valley perspective. The sales people often (not always) ask for windows laptops, but that's it. Everyone else is Mac or Linux. Servers are 100% Linux in the last decade, before that it was either Linux or Solaris. I've never encountered a windows server in 25+ years in the field and never used it myself for anything other than gaming.

It is hubris. Google thinks it can get away with anything, since they're the market leader. The reality is that a certain number of people will migrate to Firefox over AdBlockers not working well anymore, and will give FF a second life. And those are the people who helped get Chrome off the ground by installing on the computers of relatives and recommending to friends.

This, 1000 times. There is only one thing a monopoly holder is afraid of - losing the monopoly.

Im not sure Google understands they have and profit from a monopoly.

When the EU sued Microsoft, the reaction inside and outside differed wildly. Outside, there was a 'finally' feeling. Inside, Microsoftees went from denial straight to shock. They just could not understand they had become the bad guys.

They still lived in the 80s, where Microsoft was hailed as the saviour from the IBM monopoly. They might admit some individual trouble spots, but did not see these as connected to a bigger problem or core company culture. Microsoft believed their detractors were a small deluded noisy minority, but surely not the common opinion.

Google is in for a rude awakening.

I don't think that's the underlying cause. The Google product teams I've interacted with didn't seem like they cared they were a market leader. Aside from "In the making a good product" sense.

They literally just didn't understand what people not-them were doing and refused to consider it.

I remember I ditched Chrome ~2014 because I had an issue with the default download handling. Maybe auto-saving/non-standard location?

I dug around and found a bug report on it, that had been open for multiple years, with multiple Google engineers looking and commenting on it, and arguing that users would never need a config option to do... what multiple users for multiple years were asking for.

> One thing that I'm sure about V3 is that it isn't well thought-out at all.

that's because it's been invented for the sole purpose of making "breaking effective ad-blocking" look legitimate

And similarly, FLoC was invented for the sole purpose of making targeted advertising look legitimate.

There was also a lot of futile searching for use-cases for Signed Exchange spec, which was just AMP letting Google host and monitor other pages' traffic.

From what I've seen the main purpose is to increase secruity, privacy, and performance of the extension ecosystem. I have seen no sign of mv3 being for breaking ad blocking. MV3 and other parts of chromium have added features that help ad blockers.

That's lie pushed by Google. MV3 removes the ability to block requests, but keeps the ability to observe requests. Great for ads-related spying.

MV3 is more than declarativeNetRequest and the reason why declarativeNetRequest improves privacy is that users are not giving out as overly broad permission. A user may want to block ads, but may also want to not give the ad blocker the ability to observe the requests. Even if the user trusts the developer of the adblocker, there is always the chance they get hacked.

The average user should not be giving out the webRequest enabled for every website to an extension.

It makes sense for the chromium team to periodically evaluate the use cases of extensions and see if there are better APIs that it could offer to move the ecosystem forward.

If that was the case, mv3 would let extensions block arbitrary requests on a subdomain, e.g. "uBO wants to block requests on *.com, Yes/No".

Trying to keep webRequestBlocking around would mean that the performance benefits of declareNetRequest wouldn't propogate an well as it could. The worst offendors of being slow to block requests are likely to be the same extentions who wouldn't migrate to declareNetRequest.

you know what's even slower?

pulling down 15mb of bullshit ads on every single page

There are no performance benefits of declareNetRequest. One regex is faster than matching against tens of thousands of urls.

>has a 1MB storage limit

They changed this a couple days ago to be 10 MB with Chrome 112.

The 1MB limit was ridiculous. My MV3 extension uses storage.local (on disk) for session data, only because the quota is 5X bigger than storage.session (in ram). I'm waiting for 112 to hit stable so I can migrate to storage.session only.

MV3 also has this bug where extensions just stop receiving events at random. Search `lostEventsWatchdog` for a workaround.

Are you talking about local storage? If so the limit is 5 MB: https://developer.chrome.com/docs/extensions/reference/stora...

Just raised to 10 MB?

Reply on news.ycombinator.com