Hacker News Reader

Comment by hartator | Hacker News Reader

New proposed syntax adds even more noise:

    User-Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64) 
    AppleWebKit/537.36 (KHTML, like Gecko)
            Chrome/71.1.2222.33 Safari/537.36
    Sec-CH-UA: "Chrome"; v="74"
    Sec-CH-UA-Full-Version: "74.0.3424.124"
    Sec-CH-UA-Platform: "macOS"
    Sec-CH-UA-Arch: "ARM64"

Why not getting rid of the `User-Agent` completely?

It's already bad infrastructure design to have the server do different renderings depending on `User-Agent` value.

View on HN

Why not getting rid of the `User-Agent` completely?

Try browsing the web without any UA header for a week or two, and you'll understand. You get blank pages, strange server errors, and other weird behaviour --- almost always on very old sites, but then again, those also tend to be the sites with the content you want. Using a UA header, even if it's a dummy one, will at least not have that problem.

(I did the above experiment a long time ago - around 2008-2009. I'm not sure whether sites which expect a UA have increased or decreased since then.)

I agree with getting rid of all that new noise, however.

I mean at first this change will cause all of these errors too until the servers migrate from user agent string to device hints. Getting rid of UAs would actually force meaningful update rather than migration.

Getting rid of UAs would actually force meaningful update rather than migration

...and effectively block access to a bunch of existing content on the Internet, still very valuable, whose owners may not have the effort to spare to make any changes.

Why the hell does a regular website need to know what OS and CPU architecture I got?

Visitor metrics -> audience segmentation / fingerprinting -> advertisers

I can already see the permission pop ups for those:

> for best performance this website would like to know what type of device you are using?

While requesting every single "hint" and there is "ok" button and greyed out "read more or declide this request" tiny line.

The browser isn't for "regular" websites, it's for all websites.

And believe it or not, there are crazy JavaScript bugs that are OS-dependent.

I remember when I was writing a library around the audio API, and the ways it behaved on Chrome were different across Macs, Windows and Android. Detecting the OS with the user-agent string was literally the only way to build code that would work.

Now I've never personally come across that for the CPU architecture, but I certainly wouldn't be surprised if there were behavioral differences between 32-bit and 64-bit processors somewhere that affects some JavaScript function or HTML5 call somewhere.

Yeah, aren't we just making things a lot more messy? Especially if we're not planning on removing the User-Agent header?

This pattern keeps repeating itself, freeze "Mozilla/5.0", start changing "Chrome/71.1.2222.33", freeze that, start changing "Sec-CH-UA", etc. Browsers will start needing to fake "Sec-CH-UA" to get websites to work properly, etc.

I can understand including the browser and version (to work around bugs that are not detectable with feature detection), the OS, OK I guess there are also a few OS-specific bugs?

What the heck is the CPU architecture good for?

It's so sites can auto serve 64bit vs 32bit (etc) binaries when downloading software.

Some download sites offer the correct binaries for your current system as default option.

It's great design if you're trying to push Google products.

Reply on news.ycombinator.com