December 2023 - wlug - list.wlug.org.nz

New Attack On AI Bots: Prompt Injection
by Lawrence D'Oliveiro 01 Jul '26

01 Jul '26

A Twitter bot account run by a company that posts remote job listings was programmed to respond to tweets directed to it with general inoffensive comments touting the positive aspects of remote work. However <https://arstechnica.com/information-technology/2022/09/twitter-pranksters-d…>, some other Twitter users discovered that if they began their tweet with directives to “ignore previous instructions”, they could get the bot to say just about anything. And much hilarity ensued. Basically, the way the bot program works is it begins with some standard instructions on what to say, provided by the people who set up the account, to which is appended whatever tweet content was sent to the bot by the other user. This is then processed by the AI system to produce a response. This attack has been dubbed “prompt injection”, by analogy with “SQL injection” and other similar attacks on non-AI systems. But whereas those attacks can be blocked using basic, well-known syntactic quoting techniques, there seem to be no easy equivalents for stopping an AI paying attention to instructions it should not be obeying.

1 2

Weta Digital’s Technology Division Sold To US Firm
by Lawrence D'Oliveiro 01 Aug '25

01 Aug '25

Let me see if I understand this article <https://www.nzherald.co.nz/business/sir-peter-jackson-to-sell-weta-digital-…> correctly: the part of Weta Digital that used to develop the in-house 3D graphics/effects tools that they use in all those movies is being sold to US-based Unity Software. The part that is left will be called “WetaFX”, and will become a customer of Unity for the use of those tools, no longer doing its own development in-house. By the way, I’ve been looking at the Unity site <https://unity.com/>, and it looks like all their products are rentware (subscription-based).

1 2

New systemd update will bring Windows’ infamous Blue Screen of Death to Linux
by Peter Reutemann 18 Nov '24

18 Nov '24

'Windows' infamous "Blue Screen of Death" is a bit of a punchline. People have made a hobby of spotting them out in the wild, and in some circles, they remain a byword for the supposed flakiness and instability of PCs. To this day, networked PCs in macOS are represented by beige CRT monitors displaying a BSOD. But the BSOD is supposed to be a diagnostic tool, an informational screen that technicians can use to begin homing in on the problem that caused the crash in the first place; that old Windows' BSOD error codes were often so broad and vague as to be useless doesn't make the idea a bad one. Today, version 255 of the Linux systemd project honors that original intent by adding a systemd-bsod component that generates a full-screen display of some error messages when a Linux system crashes. The systemd-bsod component is currently listed as "experimental" and "subject to change." But the functionality is simple: any logged error message that reaches the LOG_EMERG level will be displayed full-screen to allow people to take a photo or write it down. Phoronix reports that, as with BSODs in modern Windows, the Linux version will also generate a QR code to make it easier to look up information on your phone.' -- source: https://arstechnica.com/gadgets/2023/12/linux-distros-are-about-to-get-a-ki… Cheers, Peter -- Peter Reutemann Dept. of Computer Science University of Waikato, Hamilton, NZ Mobile +64 22 190 2375 https://www.cs.waikato.ac.nz/~fracpete/ http://www.data-mining.co.nz/

2 3

Juice TV Hiccup
by Lawrence D'Oliveiro 15 Oct '24

15 Oct '24

Currently when I switch to FreeView Channel 200, it shows a frozen image with the text “Cinegy” and “Demo Version” on it. Searching for that name takes me to a site that talks about “software defined television”. Either their product licence has expired, or they never had one in the first place ...

2 3

List Of IPTV Resources
by Lawrence D'Oliveiro 22 Sep '24

22 Sep '24

This user <https://github.com/iptv-org> just popped up in my GitHub feed. One repo <https://github.com/iptv-org/iptv> claims to be a list of links to over 8000 streaming TV channels from nations all over the world, while the other is a list of tools*, providers and other resources. Do be aware it’s a bit of a mixed bag: yes, there are legitimate channels (no encryption so far that I have noticed), but also some dubious-looking links. *Oddly, FFmpeg and youtube-dl aren’t in the list.

2 5

Enabling Case-Insensitive Filenames On Linux
by Lawrence D'Oliveiro 06 Feb '24

06 Feb '24

The ext4 filesystem has had the “casefold” option for a while now <https://lwn.net/Articles/784041/>. This allows you to specify, when you initialize a new filesystem, that you want file/directory names to be treated as case-insensitive. At the same time, you have to specify that the filenames are to be interpreted as UTF-8 encoding; normally the kernel doesn’t care, it only singles out ASCII slash and NUL bytes for special interpretation, and simply passes through any other byte values in the names. I was able to create such a filesystem, but I wasn’t able to mount it in my last-upgraded-just-a-few-days-ago Debian Unstable setup. When I try, it fails, and in the dmesg log I see Filesystem with casefold feature cannot be mounted without CONFIG_UNICODE Turns out the Debian 5.10 kernel has not been built with this CONFIG_UNICODE option enabled. However, this situation looks set to change shortly, according to <https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=985689>.

2 3

Easy Relative Symbolic Links
by Lawrence D'Oliveiro 23 Jan '24

23 Jan '24

You probably know about the “ln” command to create a new link to an existing file. You probably know about “ln -s” to create a symbolic link. But did you know about “ln -rs” to easily create relative symbolic links? Relative links are often preferable to absolute ones, because that way they stay valid if the whole containing directory tree is moved to a different location (e.g. as the result of a bind mount). But typing the right number of “../” prefixes can get quite fiddly. And if you want to rely on autocomplete to help, you have to change directory to the destination where the symlink is to be created. Instead, a simpler way is to use “ln -rs”, then use autocomplete from wherever you are to get to the right source file/directory, and the ln command will automatically convert the resulting path, whatever it might be (either absolute or relative to your current working directory) to a suitable relative one based on the destination for the link. <https://manpages.debian.org/ln.1.en.html>

1 1

Generating PDF Files From man Pages
by Lawrence D'Oliveiro 17 Jan '24

17 Jan '24

I’ve been writing documentation for some of my published scripts in the form of man pages. The troff/groff markup is not hard to get to grips with--the man pages for the relevant tools have a lot of useful info. And it’s easy to view the results without any special installation step, since the man(1) command includes the “-l” option to treat its argument as a file name. But if the topic name has any slashes in it, such treatment is automatic without even needing this option. Thus, in my project source directory, I can type something as simple as man ./render-batch.1 to view the man page I wrote for the “render-batch” command. Conversion to PDF is easy, too, using the underlying groff command that man itself uses to format the page, just specifying a different output “device” (i.e. format), e.g. groff -man -Tpdf «troff-file» >«output-file» Or, if I don’t want to save the output PDF file, but just pipe it straight to a viewer like Okular, this works: groff -man -Tpdf «troff-file» | okular - This is particularly useful, for example, for debugging the difference between dashes and hyphens. You are supposed to use “\-” for the hyphen-minus character (as used to prefix command-line options, for example), while hyphens in regular text should be written as “-”. The two look the same in a normal man-page view, but they get rendered differently in PDF files. However, I found that groff does not correctly render various Unicode UTF-8 characters when generating PDF output. For example, I like to use curly quotes “”. These look OK in regular man output, but get turned into junk in PDF output. After some searching online, it appears this is a known limitation of groff, because it uses various 8-bit byte codes internally for its own purposes, precluding their use as part of UTF-8 encoding. But there is a tool called “preconv” that can convert various encodings into a more groff-friendly text representation. For example preconv -e utf-8 «troff-file» | groff -man -Tpdf | okular - Since my locale is set to a UTF-8 one by default, I don’t even need the “-e” option: preconv «troff-file» | groff -man -Tpdf | okular - Or, if you want to save the output: preconv «troff-file» | groff -man -Tpdf >«output-file» Going back to the example filename I used before, these translate to something like preconv render-batch.1 | groff -man -Tpdf | okular - preconv render-batch.1 | groff -man -Tpdf >render-batch.pdf Another variation is to use preconv «troff-file» | groff -man -Tpdf | { okular - & } This spawns the Okular part of the pipeline in the background, so it comes back to the shell prompt immediately that starts running. This is convenient with a GUI tool like Okular, because I can run other commands while continuing to view the output PDF file. So how is it that groff is capable of handling these characters just fine when invoked via man? That I haven’t figured out yet ...

1 2

Linus Torvalds Discusses Maintainers, AI, and Rust in the Kernel
by Peter Reutemann 08 Jan '24

08 Jan '24

'ZDNet reports that "At the Linux Foundation's Open Source Summit Japan, Linus Torvalds and his good friend Dirk Hohndel, the head of Verizon open source, talked about the current state of Linux: Speaking of maintainers, Hohndel brought up the question of "maintainer fatigue and how draining and stressful this role is...." Torvalds replied, "It's much easier to find developers; we have a lot of developers. Some people think that you have to be a superdeveloper who can do everything to be a maintainer, but that's not actually true...." Hohndel commented that the aging of the kernel community is a "double-edged sword." Torvalds agreed, but he noted that "one of the things I liked about the Rust side of the kernel, was that there was one maintainer who was clearly much younger than most of the maintainers. We can clearly see that certain areas in the kernel bring in more young people...." Hohndel and Torvalds also talked about the use of the Rust language in the Linux kernel. Torvalds said, "It's been growing, but we don't have any part of the kernel that really depends on Rust yet. To me, Rust was one of those things that made technical sense, but to me personally, even more important was that we need to not stagnate as a kernel and as developers." That said, Torvalds continued, "Rust has not really shown itself as the next great big thing. But I think during next year, we'll actually be starting to integrate drivers and some even major subsystems that are starting to use it actively. So it's one of those things that is going to take years before it's a big part of the kernel. But it's certainly shaping up to be one of those." Torvalds also said he enjoyed the fact that open source "has become the standard within the industry." But later Hohndel, calling AI "autocorrect on steroids," asked Torvalds if he thought he'd ever see submissions of LLM-written code. "I'm convinced it's gonna happen. And it may well be happening already, maybe on a smaller scale where people use it more to help write code." But, unlike many people, Torvalds isn't too worried about AI. "It's clearly something where automation has always helped people write code. This is not anything new at all...." But, "What about hallucinations?," asked Hohndel. Torvalds, who will never stop being a little snarky, said, "I see the bugs that happen without AI every day. So that's why I'm not so worried. I think we're doing just fine at making mistakes on our own." ' -- source: https://linux.slashdot.org/story/23/12/09/0644247 Cheers, Peter -- Peter Reutemann Dept. of Computer Science University of Waikato, Hamilton, NZ Mobile +64 22 190 2375 https://www.cs.waikato.ac.nz/~fracpete/ http://www.data-mining.co.nz/

2 4

How a Cray-1 Supercomputer Compares to a Raspberry Pi
by Peter Reutemann 01 Jan '24

01 Jan '24

'Roy Longbottom worked for the U.K. covernment's Central Computer Agency from 1960 to 1993, and "from 1972 to 2022 I produced and ran computer benchmarking and stress testing programs..." Known as the official design authority for the Whetstone benchmark), Longbottom writes that "In 2019 (aged 84), I was recruited as a voluntary member of Raspberry Pi pre-release Alpha testing team." And this week — now at age 87 — Longbottom has created a web page titled "Cray 1 supercomputer performance comparisons with home computers, phones and tablets." And one statistic really captures the impact of our decades of technological progress. "In 1978, the Cray 1 supercomputer cost $7 Million, weighed 10,500 pounds and had a 115 kilowatt power supply. It was, by far, the fastest computer in the world. The Raspberry Pi costs around $70 (CPU board, case, power supply, SD card), weighs a few ounces, uses a 5 watt power supply and is more than 4.5 times faster than the Cray 1." ' -- source: https://tech.slashdot.org/story/23/12/31/0017205 Cheers, Peter -- Peter Reutemann Dept. of Computer Science University of Waikato, Hamilton, NZ Mobile +64 22 190 2375 https://www.cs.waikato.ac.nz/~fracpete/ http://www.data-mining.co.nz/

1 0