I hadn't ever heard of that before, so I looked it up. Can you elaborate as to why/how you used it? Even after reading about it, I can't see any instance where I'd need something like that, so I'm just curious.

Also, with regards to Xenial and anything newer crashing, some versions of Ubuntu come with "HWE" (Hardware Enablement Stack). I think it's just in the point releases, like 18.04.1 but not in 18.04. I have no idea if that will help in your case - but it *might*, so I figure I'll bring it to your attention.

If memory serves, back in the day we had to configure it ourselves - or at least I recall having to. It was many years ago and I think I needed it for a keyboard or mouse layout? I might be thinking of something else entirely.

No, that was Xorg. It scared the life out of me because I had no idea how a video card actually worked or what a ramdac clock was or modes (I still don't know). Fortunately, you could run Xorg with a configure argument and it would create a default Xorg.conf file that would work for most systems. You just needed to edit in things like the keyboard layout.

One thing I learned a couple of days ago is that I had it wrong about directory permissions. I thought you needed read access to access files in the directory and execute access to attach and list. Someone told me it was the other way around. Execute access is what you need right down the path to your file, but you only need read access if you don't know the filename.

Right! Thanks! Your memory is better than mine. That leads into this:

I see you use LFS and that's awesome.

I don't - as it'd be so much more work than just a quick and easy install with something that has a handy GUI to do it for me. I really, really appreciate the ease of doing so. For me, it's okay because I don't need anything complicated and the defaults match what I'd generally end up doing myself.

It's why I don't (generally) use Gentoo, Slackware, or even Arch. I have VMs of the first two (no VM for Arch since moving some stuff around) and they're fun to go through and setup, but I hose my OS frequently enough to value the simple install routines with a nice cushy GUI.

I played with my first programmable computer in 1971 or '72. I didn't like it. I'd try again in the late 70s, early 80s. I still didn't like it much. It wasn't until the mid 2000s, when PCs became really easy to use _and_ had enough resources to not be bogged down all the time, that I really started to appreciate them. It was very much a love/hate relationship prior to that. I legit like simplicity, though I suppose we all have different views of what's simple.

iPXE is just a more powerful way to do pxe boot stuff. In my case it is the default choice for the new LTSP software. I figured stick with what they use instead of giving myself more problems. It gives me more options as well.

Google "radeon ring 3 stalled for more than" and you will find a bunch. I looked many of them. I tried every suggestion I could find. I did find a bunch of debating on where the problem exists though. Some seem to think mesa, others the radeon driver, others the kernel, others X, and of course some architecture flaw in the gpu itself. It seems to only affect 1 generation of AMD gpus though which I happen to have in 2 of my kodi machines. As it is an older generation it will probably never get fixed, if they ever even find it. I've had the issue on 18.04, and 20.04 when I tried it yesterday.

Like I said though fortunately these are nothing more than Kodi portals, no desktop use. As such they already are fairly isolated and we only run local media off of NFS. No streaming on them. I simply marked the current Kodi installations with apt-mark hold on them, and on my 20.04 that I have on my desktop and living room tv / desktop and that should keep Kodi static and not cause any DB issues when they update Kodi if they don't include Xenial. Not ideal of course. If I had the cash I'd just get different hardware. But it's hard to justify that when they work perfectly fine as they sit now. By the time I'm forced to move on in Kodi versions I figure I can more justify the hardware upgrade. 5 years down the road maybe, when 20.04 goes eol.

I've always been less concerned with immediate ease of use than with the ability to understand what I'm doing. After all, you can learn how to use something, especially when you're retired and have plenty of time on your hands. I'm basically tech-phobic, so I hate the thought of computers doing things behind my back. LFS allowed me to create a very minimalist system which contained the applications I needed but was still comprehensible. Unfortunately with the modern network of dependencies, that's no longer possible, so I don't really use LFS any more. Slackware has become my goto distro. But I still build the new LFS version when it comes out, just to prove to myself that I still can.

Gentoo and Arch are both bleeding edge distros, which I don't like. And Arch has a very strong community with its own ways of doing things, which I find oppressive.

You often learn things while answering someone else's questions. I just learned something new about /etc/mtab.

In most distros (including LFS), this is just a link to /proc/self/mounts, but in Slackware, it's a separate text file. This intrigued me, so I found out which startup script created it (rc.S) and looked at the code. It's quite straightforward, writing in just the root partition and the dynamic ones. Anything else gets added later. But when I compared the contents of this file with what /proc reports, I could see at once why Patrick did it this way. The file in /proc contains all kinds of rubbish from /sys/fs/cgroup, whatever that is! The Slackware mtab file just shows you what real partitions are currently mounted, which is surely what you want to know when you look at this file. Now that's what I call user-friendly!

But when /etc/mtab is a text file, how do new mounts get into it? Simple! According to the mount man page, the mount and umount commands check for a real mtab file (i.e. not a link) and edit it as required.

That actually makes a lot of sense. I'm more of an 'immediacy' and 'convenient' kind of person with regards to computing. In days past, I used to even build my own computers (from parts, not literally fabbing ICs!). These days, I buy something made - usually aiming for something that has been out for a year so that it probably has Linux drivers, and call it good.

I remember the old days with cradle MODEMs (acoustic couplers) with the AT commands and even needing different settings to connect to different BBSes and stuff like that. I am more or less glad those days are over.

Then again, I think the polite way to say this is, "The Internet was a lot different when it required a degree of skill to access it."

I'm feeling lethargic today. It's a long story.

So, let's just add to this thread rather than starting a new one.

------------------

In the past seven days, I've learned a great deal. I learn a lot every week!

This week, I went through a complete MySQL install just so I could write directions about how to remove it. It has its tendrils everywhere.

I also installed Wine just to help someone else and it's *insane* how much data it pulls in and how many dependencies it pulls in. It's insane. I'm glad I don't need to use any Windows software. I don't even have Wine installed on this computer - though it is in a couple of VMs.

Speaking of Windows software, I do miss "Crimson Editor" from Windows. IMO, best text editor ever.

Anyhow, MySQL is a pain in the butt to remove all traces.

In Gimp 2.8 I learned how to make a YouTube channel 'Image,' to b inserted into the video on the bottom, by using three layers. Learned not only the how to create the three layers but also using the 'Ellipse' tool (which I had never used before in layers).

The exercise also helped me to understand more about the usefulness of Gimp and its potential.

Not sure I "learned" so much, but I spent some time with SailfishOS (mostly struggling to get a custom script working, not part of that article).

And I learned how to change xterm's fugly default icon!

Oh, and I dreamt about coding last night.
Not entirely, I didn't actually dream about sitting at a computer writing the script, but I was thinking it up as the dream went along, very real bash scripting with user input and quality control. While taking the company laptop on a cumfy train journey to Switzerland.
I have a very real idea for a small-scale flat file user info database now, if need be...

Is that your site?

Learned about cpu-x . Subs for cpu-z like in Windows.

Aint gonna use it. lshw and inxi/pinxi work Ok for me

Word.

I learned how to bind my server's media folders into my smartphone's filesystem with sshfs!

For the past 3 weeks I have been struggling to recover data from a failing hard drive belonging to my Brother who uses Win 10. Although I have not once had a problem with Linux ntfs or ntfs-3g, I did assume that Win 10 would be better suited to repair and recover from an ntfs drive. When I couldn't get Windows to even mount the drive, spinning for hours, I tried Slackware. It came up immediately in Dolphin and I copied about half of the data in the first pass.

After that, fsck from CLI repaired the disk enough that it would actually boot (well...after almost 7 minutes to Deesktop) and I could force a deep chkdsk and defrag. I still couldn't recover 100% of the disk so I tried a Knoppix by Runtime bootable CD. I got over 96% of the data recovered and it would then boot to desktop in about 3 1/2 minutes. I backed it up and sent both drives back to my ever so grateful Brother.