Wonder what the chances are that her voice was deep faked. I don't want to get all conspiracy theory and say it probably was. But considering who's involved, the chances are definitely non-zero. Feel like this is obvious neferious use case for the tech. Impersonate loved ones to manipulate.

Off the top of my head, offsceeen canvas and being able to query for many processors the client device has to use for web workers

Used power cables for my Thinkpad from a computer recycler. Have them all over house and dedicated one for my bag so I never think twice about packing or forgetting it somewhere

As someone getting into front end dev with WebGL I constantly want to do this but with a message to specifically call out Safari. Just yesterday I found out they don't support navigator.hardwareConcurrency

My creativity and explorations shouldn't be limited by one vendors intentional gimping of the web to shield their app store revenue. And calling for devs to not push things where possible but instead go at the pace of the slowest player feels very "No Child Left Behind", which also leaves a bad taste in my mouth.

It was supported for a while but just ended up being more free bits for fingerprinting, Firefox spoofs it if resistFingerPrinting is on for example.

I have no proof, but I feel like it's more than that. I don't think getting number of threads is fine grained enough to make a meaningful difference on identifying a device. Especially compared to fine grained numbers I've seen extracted doing audio fingerprinting with AudioContext. At least with firefox I can give the user instructions on how to enable it again if they want to and they can make the decision for themselves. From what I read briefly last night about safari, it's impossible to get a number directly unless you wanted to recompile the browser yourself. I found this package (https://github.com/oftn-oswg/core-estimator) which seems to give an estimate, but its something someone who wants to fingerprint could also implement on a library level, so we're back to square one except with worse developer and customer experience.

Offscreen canvas is also something I wish more browsers supported, along with webgl2 (need for fbo). Although, I've heard rumors webgl2 might be finally coming to safari soon.

I also wouldn't have an issue with safari taking hard stances on these things if only they allowed competing browser engines on their phones.

Could probably roll something yourself using Spotify API. Has tempo available as a query https://developer.spotify.com/documentation/web-api/referenc...

This exists. I use it daily & it is awesome. Its called spotifyd: https://github.com/Spotifyd/spotifyd

I don't really understand how this fits in.

Oh man Thank You. I came across this video (https://youtu.be/rpkoCZ4vBSI) years ago and I've haven't been able to find it again since. Your comment gave me enough of a breadcrumb to finally track it down

Thank you

Being able to select your default browser doesn't address the issue that every browser on iOS is just a webkit wrapper

I found a lot of very valid points presented in the second half from gesture navigation to inefficiencies in default navigation I hadn't thought of. Watching at like 2x speed is a must tho. I wouldn't have been able to make it thru otherwise.

Does anyone know if its theoretically possible to run a cluster of Pi's off of the same external ssd at the same time? I'd like for say 4 Pi's to get a speed boost of not running off microSD, but I also don't want to shell out for 4 separate hard drives.

I've looked in the past and have only found links to partitioning the drive so that it could be used when moving from one pi to another. But nothing about using it concurrently.

I'm not even sure what this would look like physically. Like if a hub exists that could pass traffic to a single drive connected on the other side. Or how the drive would internally separate the writes. But im curious if something like this has been done before. Or why it couldn't be done

PXE Boot w/ NFS root - you won't get the bandwidth (about 112MB/s over gigabit Ethernet), but you'll definitely see the response time improvements (IOPS + latency).

This. I intend to do some more testing with my current cluster to see how much improvement I can get with one Pi 4 serving the traffic (vs each Pi running from its microSD card). And ideally seeing if a faster machine with faster storage could do even better.

If you end up doing this you should definitely make a post about it. I'm sure I'm not the only one incredibly curious on the real world performance and tradeoffs

I'm not aware of any solution that exists to connect a single USB device to multiple hosts simultaneously.

You need a 5th PI in front of the SSD providing a network file service.

Networked storage but that won't be as fast as native drives.

I tend to disagree. It will depend on the network and the storage but it is not uncommon for network storage to actually be faster than local flash card.

At home I have couple of older laptops my kids use for various things. I have installed linux on them and configured to boot over wifi and it actually sped up these machines significantly. I would not advise for everybody to do this, though. The setup is complex (there is no built in support in any distro and you are pretty much on your own) and you need excellent wifi coverage (I have no less than 5 5GHz access points to cover the area densely).

One thing flash card will be difficult to beat is random small accesses if the only thing you care about is latency and not throughput.

It's a gigabit network adapter. Ignoring the extra CPU overhead, latency, etc. that alone limits storage to less than SATA 1 speeds. UASP drives on the USB 3 port can push near SATA 3 speeds with lower latency and less CPU overhead.