"This story sounds a bit odd"
FWIW we're reporting what the auditor said - so if something looks odd, you mean, the auditor's findings are odd.
C.
3532 publicly visible posts • joined 21 Sep 2011
"surely that data should have gone directly to a NASA data centre first anyway"
That's the rub. NASA didn't want to run its own data centers: it opted to upload all the stuff gradually to the cloud. According to the audit report, it didn't realize that people can't download this stuff "for free" from the cloud – someone has to pay for the bandwidth. NASA, in this case.
C.
Thanks for the feedback - I've tweaked that sentence.
Don't forget to email corrections@theregister.co.uk if you spot anything wrong. We can't read every comment, but we can read every email to that address. Case in point: if you had emailed us, we could have addressed this hours ago.
C.
From the court document:
"At a subsequent meeting on March 15,2018, Pankowski reiterated the statement he made during the earlier meeting that his team could not approve the product because it did not meet safety requirements and doing so would place consumers of the product at a safety risk and expose iRobot to liability."
And:
"iRobot ... refused to provide required safety and labelling information with the products it sold"
Emphasis mine. Hope this helps.
C.
They are not vulnerable, it appears:
https://portal.msrc.microsoft.com/en-US/security-guidance/advisory/CVE-2020-0796
C.
"Why is it that The Reg publish the tiniest grainy photos of products, when a full resolution copy exists?"
Because we sometimes forget to link to the full-res - and forget that you can't click on the main top image to expand it.
FWIW our articles - the content written by journalists - are input and edited as raw HTML in our publishing system, which then sticks stuff like the mast-head, headline, comments link, and so on, automatically when the article is published.
The top image of articles is automatically inserted: we can control the URL of the image, and whether one appears, but it's always non-clickable. If we want to make the main image clickable, we have to manually hide it from the top and embed it lower down in the piece.
In other words: if there's no link to a larger version of an image, someone forgot to put the anchor tag in. If the top image isn't clickable, and should be, someone forgot to move it down into the article.
Drop us an email to corrections@theregister.co.uk if you think we've screwed up so we can fix it immediately.
I've moved the top image down into the article and made it clickable.
C.
You need to be a black-belt in reoccurring inside jokes to work around here.
It's partly to annoy pedants, and partly a send-up of publications (including El Reg TBF) that accidentally mix up pictures of TV shows, airplanes, battleships, old computers, actors, etc, in headlines and images.
C.
I think it was more checking they were operational.
FWIW, ICANN has the ability to override protections and literally drill its way into accessing the KSK HSM but it's rather obvious if that were to happen.
The point being that IANA/ICANN staff can check security systems but there are tamper-proof protections and other layers to prevent actual access outside of a ceremony, unless you brute force your way in, which is, shall we say, detectable.
C.
It's fairly trivial facial recognition - it's the scale and the source of the training data that's causing people to kick off.
Here's how I'd do it. You take 3 billion pairs of images scraped from online profiles and URLs to those profiles. You train a convolutional neural network – or a series of networks – to map images to their source profiles. To make life easier, assign each profile an ID number. Thus a particular face will map to ID 1000, another to ID 1001, and so on.
So when you show it a face, it predicts what the correct profile ID should be. Thus if you show it an image it hasn't seen before, it will try to map it to the closest matching face and its profile ID. You then turn that ID number into a profile. You now have a suggested identity for that input face.
The neural network can output profile ID numbers with a confidence value, so a face could return ID 1001 90% confidence, ID 3000 70% confidence, ID 2000 10% confidence, etc. Just take the highest two or three.
Depending on the training and input data preparation, the training process and the network architecture, it'll be accurate or not very accurate.
As for the scraping: buy a lot of cloud instances and parallelize your curl fetches, crawling webpages, building a graph network.
C.
PS: Cache a copy of the page per URL so that if profiles disappear online, you still have a copy. These pages will contain stuff like names, personal info, links to other profiles owned by the person, etc
1. We like to think it's obvious to Reg readers that 75% (it's actually 77) is low. Didn't seem worth making too much hay about it.
2. Kieren wrote 3 great articles on Tuesday, and we at the back edited a load more. We're always pushed for time, there are always improvements and extra work that can be done, and we have to ship a product at some point. Sometimes you have to call it and run it, and move onto the next thing that needs covering.
C.
I've heard rumblings that a chunk of the data center sales are or were due to banks and cloud builders replacing processors with those with mitigations built in.
But it's mainly Intel auctioning off chips to hyperscalers ordering 1m parts a year, I reckon.
C.
The original complaint is way out of date - the legal battle has changed over the many years it's been running. Various claims have changed, or been thrown out.
I recommend reading the latest appeals court ruling: https://regmedia.co.uk/2018/03/29/oracle_v_google_opinion.pdf
The original is here: https://regmedia.co.uk/2010/08/13/oracle_complaint_against_google.pdf
C.
For bare metal, you can use what's called the core library, and a few others, that provide primitives and basic structures. There's no IO nor heap allocator: you have to implement that yourself.
If you pull in a third-party library that expects to use std library primitives, you're out of luck in a core-only environment. It's not the end of the world: not all libraries require std. Ones labeled nostd will work with core.
std expects there to be an OS underneath providing stuff like heap allocation and IO. With bare metal, there is no OS. You are the OS. Core doesn't require an OS. Core expects you to provide things like the heap allocator.
So, with core, you're not totally on your own, you get some support, you just can't use dependencies that require std, and you have to provide stuff like memory management.
Compare std: https://doc.rust-lang.org/std/
to core: https://doc.rust-lang.org/core/
Core gives you a lot, along with common data structures in the alloc library, but not all of std.
C.
FWIW you still have to obey Rust's borrower rules in unsafe code. You also wrap unsafe { } around unsafe code so you can find and audit it easier - whereas with C/C++ any old pointer could be wild.
Unsafe IO port access on Rust looks like this:
unsafe { *(0x10000000 as *mut u8) = c };
..for writing a character to a serial port at 0x10000000, say.
I've done a fair amount of low-level Rust dev in my spare time and it's helped me write much better code, so much so, I am terrified about the thousands of lines of C I've previously written.
I'm not on the Kool Aid, though. It's frustrating that a lot of the ecosystem uses the standard (std) library, which is unavailable on bare metal. Thus, you may find a third-party library that will save you reinventing the wheel, only to find it's reliant on std and you can't use it.
C.