The Technology Vent and Rant Thread

[madelonw]

I'm beginning to despise Amazon all over again.

I am in the process of downloading my extensive Kindle library (a bit OCD about my books, don't trust Amazon to not f*** up) again! Anyway, I've gone through a couple of steps to get where I am in the process. I use the latest version of Kindle for Windows which I have made my default device. Next, I use Epubor to convert the downloaded book to EPUB. I add those EPUBs to Calibre. (That's the short version.) After trying to do the downloads directly in Kindle for Windows (search, mark, download), I thought a better way to do it was to use Content and Devices (search, send to device) which seemed to be working. Now, I find that Select All and Deliver to aren't exactly working (I look for "in 1 Device to confirm that books were delivered. And, for the past couple of days, I am getting a message when I try to select a page of books and try to send them to my default device: Request did not process. Please try performing the operation again. Where the bulk download is giving me grief, I can download the books one at a time or in small (3 book) batches.

I don't know where to post this (other than here). Any advice? I'd like to know why this is happening.

[Cactus Chef]

Originally Posted by madelonw

I'm beginning to despise Amazon all over again.

I am in the process of downloading my extensive Kindle library (a bit OCD about my books, don't trust Amazon to not f*** up) again! Anyway, I've gone through a couple of steps to get where I am in the process. I use the latest version of Kindle for Windows which I have made my default device. Next, I use Epubor to convert the downloaded book to EPUB. I add those EPUBs to Calibre. (That's the short version.) After trying to do the downloads directly in Kindle for Windows (search, mark, download), I thought a better way to do it was to use Content and Devices (search, send to device) which seemed to be working. Now, I find that Select All and Deliver to aren't exactly working (I look for "in 1 Device to confirm that books were delivered. And, for the past couple of days, I am getting a message when I try to select a page of books and try to send them to my default device: Request did not process. Please try performing the operation again. Where the bulk download is giving me grief, I can download the books one at a time or in small (3 book) batches.

I don't know where to post this (other than here). Any advice? I'd like to know why this is happening.

2009 join date and you're only doing this now? You unfortunately missed the boat on doing things the easy way: it's a very complicated endeavor now, and only gets moreso as time goes on.

Most of us read the tea-leaves when Amazon announced it was removing the D&T option and got out libraries out swiftly.

[DNSB]

Spent a chunk of yesterday thinking nasty thoughts about computer technology. For reasons that some people seem to be sufficient, British Columbia went to a permanent UTC-7 time zone. When the switch to standard time happened yesterday, it was great for the dumb clocks since nothing needed to be done. My iPhone switched and when I checked, in date & time, there was a note about rebooting to install new time zone information. After the reboot all was happy. Then I got to the Android devices. My Lenovo Smart Clock Essentials (a speaker/clock combo) and my Lenovo M8H8 tablet are now residents of Dawson City, YT which is MST (UTC-7, no DST). Perfect except the temperature display is now for a city ~2000km north of my location. It's just a bit nippier there. The Windows computers just needed the time zone set to Pacific Time and my Kobo ereaders needed the UTC offset updated.

Google will have updated timezone information for my older devices some day soon.

The only consolation is that I should not need to do this again for a few years.

[Terisa de morgan]

You can try to go from Spain (with DST) to Brazil (with no DST) and try to do your phone to behave (note: I left Brazil before being able to do it)

[bizzybody]

I've used AI to help write a full-featured GUI for Microsoft Storage Spaces that does all the stuff Microsoft's GUI can't. I've also used it to write a RS232 serial control app for late 1990's Toshiba digital projectors, and a script for MyDefrag that gives it the missing capability to Consolidate Free Space AND Defrag. Yesterday I used it to help me figure out how to get Subtitle Edit to do correct speech to text in English and Spanish of an old 30 episode TV series for teaching children Spanish.

All that and more with Google's Gemini. There are AI's with paid subscriptions that are more capable on technical stuff and can do more of the work themselves VS giving the humans instructions on how to do it on their computers.

It's an amazing tool to help people do things they know pretty much exactly *what* they want to, but don't know *how* to do them.

One quite annoying tendency Gemini has is when you tell it you're using a specific version of software like Subtitle Edit 5.4.0 it will repeatedly forget that and give directions for an older version that is quite different.

On two different projects I asked it to summarize our chats into Word files for me to download. In one project it made them quite detailed.

In the other they're the barest highlights. I also asked it to generate a PowerPoint file, which it did, but every slide just has a title and boxes with "Click here to add text" instead of the text from the chat thread. So I copied and pasted the relevant parts of the thread to a doc file. After it prepared the nearly useless Word files it suddenly became completely unable to generate downloadable files. I did get a functional HTML5 document out of it, indirectly, by copying and pasting Gemini's output to a text file.

It has adjustments for various amounts of money going to every US State. I tried for a long time to get it to create a version with options to look at every county or equivalent but it flat out was unable to make that version functional. I also had to keep telling it I wanted it for all the States but it would keep reverting to a subset between 4 and 8 States.

Sometimes it's like Gemini has the memory retention of a goldfish.

[ratinox]

Originally Posted by bizzybody

Sometimes it's like Gemini has the memory retention of a goldfish.

Gemini and other generative pre-trained transformers (GPTs) have no "memory". They don't think. They don't know anything. They're statistically weighted prediction (guessing) engines.

One quite annoying tendency Gemini has is when you tell it you're using a specific version of software like Subtitle Edit 5.4.0 it will repeatedly forget that and give directions for an older version that is quite different.

Because it's a statistically weighted guessing engine. It will randomly "forget" things, where "forget" really means "semi-randomly pick something you don't want". This is not a bug. It's the nature of GPTs. Everything they do is statistically weighted pseudo-random guessing. Sometimes they guess right.

It's an amazing tool to help people do things they know pretty much exactly *what* they want to, but don't know *how* to do them.

Exactly. Consider what it means when you don't learn how to do something you want to do. Can you honestly say you actually did these things? Or did the machine do them for you and maybe you just tidied them up? Consider what benefit you gain from letting something else do it for you, and consider what benefits the operators of that thing gain from your dependence on their tool.

[Renate]

"AI" is a mixed bag.
I don't ask it to do anything that I would do myself, like engineering or coding.
But if I have a question like, "Why should ducks not eat oil or fats but songbirds eat hard fat?", I'm content to rely on AI.
I can have a "conversation" with Gemini why Aachener printen are boring in comparison to a good molasses hermit, with reference to Russian baked goods.
The state of the art has and will improve.
Maybe some day we'll actually get real AI.

[bizzybody]

Just now, thanks to Gemini, I have a custom Debian ISO on my Windows desktop that I can write to a bootable USB drive or directly to a SSD or HDD then install in a target system.

Boot off the USB drive and install, or boot off SSD or HDD and it will ask a couple of setup questions and complete its setup.

What does it do? It's a zero configuration DLNA server. Serve media files by DLNA is all it does. Just write down some info given during setup to map it as a network drive then copy files to it.

Everything put on it is made available via DLNA on the local network. All internet communication to the server is disabled.

The super simple DLNA box I've wanted for years, and I just had to install Debian in WSL2 on Windows 11, get it up to date, prompt Gemini with exactly the features I wanted and after a bit of back and forth I got a script to copy to the Debian prompt to download all the bits and pieces it needed, then 4 pieces of script (working around its code block output size limitations) to paste at the prompt to build the ISO.

To figure out how to do all of that on my own would take a huge amount of time and a lot of trial and error. I have asked for advice on how I could put together such a dead simple DLNA server and all I ever got back was essentially "Why would you want that? You should use Open Media Vault, or Plex, or Jellyfin etc."

All of those and others are far, far too bloated. They are loaded with features and capabilities absolutely not needed to fulfill the task of simply making media files available that client devices (Smart TV, phone, laptop, PC etc) already have the built in capability to view or play. I have zero need to transcode anything on the fly.

This custom Debian build should do what I want and nothing else, though I did have it configured so that it *can* have a master login and users and passwords, with space and time quotas *if desired*.

I'm going to save the scripts and other relevant info so anyone can build the same ISO.

It just seems to be the "default" in the Linux community to ignore what someone asking questions says they want to do, and insist that you should either not do it at all or that you really should pile on a bunch of things you explicitly said you don't want.

Gemini took in what I said I wanted for my "Ideal minimalist DLNA server", figured out how to deliver exactly what I wanted, with no BS about what I really should use.

In other words it acted like a computer, did what it was told and nothing outside that.

What some of these AI's can be used as is like a general purpose expansion to the "Conversational" CNC milling machine systems that started in the 1990's. They were/are systems where an operator can enter information directly on the machine to create a design for milling something instead of having to directly write a gcode file or design in CAD software that could output a gcode file.

[ratinox]

Originally Posted by bizzybody

Gemini took in what I said I wanted for my "Ideal minimalist DLNA server", figured out how to deliver exactly what I wanted, with no BS about what I really should use.

And here I am typing "linux dlna setup" into the startpage search box and getting a list of step-by-step instructions to install and configure, and would have me up and running in less than a minute. And I'd have discovered that it's not difficult or time-consuming at all.

But you do you.

[cc_in_oh]

What I need is AI that can reconnect all my so-called "smart" devices after changing wifi passwords. And then get them to work with Alexa again...