The officially official Devuan Forum!

You are not logged in.

#1 Re: Other Issues » Update nginx > 1.19.4 » 2023-06-19 00:29:05

delgado wrote:

Looks like they don't have a devuan repository.

Nobody has ever had and probably never will have a repository for "Debian Chimaera".

Other than that, I really would not expect any third party to have Devuan repositories.

Seems like he was just throwing stuff together without understanding what any of it does.

The only real answer would have been "no, it's not going to work", not a Debian package in a Devuan system (which was not what he was trying to accomplish) and not a Bookworm / Daedalus era package in a Bullseye / Chimaera era system (which was what he was trying to accomplish).

#2 Re: Desktop and Multimedia » EXT4 BUG? Data loss when I have launched sigil :s » 2023-06-18 23:38:56

So wouldn't the correct course of action be, in this case, to close this bug report manually?

If not, then just tell me. I'm only learning this stuff.

I think we have now established that 1) he is not going to reproduce it, ever. And 2) the information provided seems to point to a problem with crypto, not with ext4, not with Sigil and not with his hardware, but 3) we can never be sure and this isn't going anywhere ever.

That is, if we think that what broke, broke essentially (the problem, whatever it was, broke everything that it could, namely all encrypted filesystems) and not accidentally (it could have broken any filesystems, but it just happened to break these particular ones this time).

We have also established that 4) there is no way for Devuan or any distro of any size to reproduce a problem like this, also based on the information given, but also because no distribution is going to start a hunt for a random problem based on one report. Even if there was widespread property damage or loss of life, there would simply be not enough information to do anything. There simply isn't anything obviously wrong with ext4, and it would be his problem to prove otherwise.

I personally feel that a distro does not look bad if it can't do anything about this. But a distro might look bad to some people in some cases if it has many critical bug reports hanging endlessly, because someone might think that there are serious problems the distro thinks it ought to fix but for some reason hasn't been able to. It muddies the waters, so to speak.

#3 Re: Hardware & System Configuration » Thinkpad T14s gen 3 and display backlight (adjustment not working) » 2023-05-28 00:14:16

Seems like it suddenly started working. Will write again if there is still a problem.

#4 Re: Off-topic » AI on mainstream media » 2023-05-28 00:12:01

LU344928 wrote:

Firstly, to anyone who tends to deride AI by claiming it can never equal human intelligence, if you subscribe to the Simulation Theory (or argument, as its author calls it) then we humans could very well be AI.

Do you have some kind of emotional attachment to AI or Simulation Theory? If not, then why do you imagine and insinuate that those who have different ideas are motivated by strong emotions? Such as derision. There is simply, plainly, obviously no reason to believe that any AI running in today's computers could do anything but remain artificial.

#5 Re: Desktop and Multimedia » EXT4 BUG? Data loss when I have launched sigil :s » 2023-05-27 23:23:49

I think I got my answer. That particular bug is never going anywhere and might as well be closed with worksforme or something like that. (Or unreproducible. That exists.)

zero wrote:

It is a fact that bigger projects like Debian, Ubuntu, Fedora, etc. handle better this kind of issue. Here an issue with Fedora and XFS:

- https://bugzilla.redhat.com/show_bug.cgi?id=2208553

The first obvious difference is that the person who reported that bug could and did reproduce it. With a report like yours, any distro, no matter how large, would have simply done the same thing Devuan did: asked you for more information.

There simply exists no way for them to reproduce your bug. What should they do? Run ext4 in random configurations and see if they encounter it? Well this is what millions of people are doing all over the world anyway. Every ext4 user is already technically hunting for your bug without even knowing it.

As far as I know, every person in this project (referring to Devuan) is a volunteer. And the average person who will respond to you on this forum is either a) doubly a volunteer (because not even appointed volunteers respond here in any official capacity, as far as I know). Or b) only a user like you. The project's response is what you got in January of 2020. And the best thing you're going to get from any developer in any corporation or project to this issue is either moreinfo or unreproducible.

If photorec could not find any of your pictures then either a) you ran it on the raw, encrypted device. Or b) you ran it on the logical, unencrypted device, but something had overwritten the files. Did you still see the decrypted devices (nvme_crypt and sde1_crypt) after the fact? If you did not then it's possibly the crypto that failed (ergo not the filesystem).

#6 Re: Desktop and Multimedia » EXT4 BUG? Data loss when I have launched sigil :s » 2023-05-24 20:24:33

Did this happen again or did someone find the cause? I mean, the bug report still exists. But what are the chances of it ever going anywhere?

I am not especially interested in this particular bug, but I have always wondered what policies distros have for hanging bug reports. Can a report be in a moreinfo state for years? Has everyone forgotten this, or is someone trying to reproduce it?

I take it that in reality there wasn't an attempt to reproduce it, neither with a tainted nor with an untainted kernel. I am sorry for the loss of data that happened, but this stands now, four years later, as an act of God, strike of a cosmic ray or a butterfly flapping its wings in Japan and causing a data loss halfway across the globe.

Bug reports by definition can't do anything for disasters that probably only ever happen once. Unless the cause is known beforehand.

If there is a known cause, I'd suggest that something happened to the crypto key. It was altered in memory somehow and some important part of the disk's data or the on-disk representation of the key got overwritten. That wouldn't explain though why two drives were lost.

Were the disks physically fine afterwards?

#7 Re: Installation » Install ZFS » 2023-05-21 01:41:58

SpongeBOB wrote:

any ideas ?

Your problem is that the modules you build need to be built for the version of the running kernel of the livecd. So you need to manually download the matching version of kernel headers and a few other things. For some reason it otherwise pulls in slightly newer versions of all the relevant stuff, builds the modules accordingly and finds out that it has nothing for your running kernel. I think you need

linux-compiler-gcc-10-x86_5.10.158-2_amd64.deb
linux-headers-5.10.0-16-amd64_5.10.127-2_amd64.deb
linux-headers-5.10.0-16-common_5.10.127-2_all.deb and
linux-kbuild-5.10_5.10.158-2_amd64.deb

or whatever uname -r tells you. Do not get confused by package version (5.10.127-2) versus kernel version (5.10.0-16), or the other way round (see, like I just did).

#8 Re: Off-topic » AI on mainstream media » 2023-05-16 23:54:40

pcalvert wrote:

I saw a short clip from a recent Elon Musk interview. In it, he mentioned that AI is already being taught to lie.

What really happens is one thing. How it is marketed is another.

I am surprised that so many people are simply technically illiterate when it comes to how an AI works. That, or I have studied psychology and odd metaphysics far more than the average person.

Intelligence or intellect literally means the ability to make sense of something in relation to some purpose. All animals have what is called passive intellect that picks out relevant things from sensory input, helped by memory and instinct. A computer does not have instincts but it does have memory, and this is what an AI basically emulates: an animal without instincts. No AI, just like no animal, is aware of any purpose, but mechanically matches its input to the contents of its memory (and instincts, if any).

Humans have what is called "possible intellect". No animal and no AI has ever had any idea what should be or what could be. Therefore an AI can never literally imitate anything, any more than a line on a paper "imitates" the path taken by a pen. No AI has any concept of "truth" or "lie", any more than Pavlov's dog "lied" when it drooled after hearing the bell, making an innocent bystander think that maybe it sees food somewhere.

The thing that an AI does impressively well is that it smuggles Aristotelian logic into what otherwise would be a meaningless jumble of words and probabilities. Not because it understands logic, because it doesn't, but because the input material is sometimes ordered accordingly.

When people used to crack passwords hashed with md5, I found it impressive that someone could throw entire dictionaries and libraries at them. If your password was any phrase from Shakespeare, Tolkien or Star Wars, even slightly modified, if the hacker came up with the same rule you did, he could reverse it almost instantly.

In these language models today there is so much capacity that the English language simply is no match for it. There is no need to feed the AI all literature ever written, because after a few thousand books there won't be that many new patterns or combinations. The sheer capacity makes it possible to emulate the "understanding" of such concepts as truth or falsehood by simply designating certain word combinations as one or the other. An AI could describe "pike is a fish" as a true statement, but you could as easily ask it if "pike is a fish" is a left or a right, up or down, fast or slow statement and get an idea of the logic involved. It could for example say that it's more of an "up" than a "down" statement. Because it operates purely by numerically weighed assosications that simply give the appearance of logic or understanding.

#9 Hardware & System Configuration » Thinkpad T14s gen 3 and display backlight (adjustment not working) » 2023-05-16 01:27:41

Standardpoodle
Replies: 2

I installed Daedalus 20230508 into a fresh Thinkpad T14s gen 3, but no method that I've tried lets me adjust the display brightness.

Directory /sys/class/backlight is empty.

I think xrandr doesn't adjust the backlight at all.

With or without sudo, xbacklight has no effect.

Adjusting the brightness works in Ubuntu 22.04 liveCD. There is /sys/class/backlight/amdgpu_bl0 that is populated.

Installing firmware-amd-graphics 20230210-5 caused the login screen to remain black (with backlight on). The directory under /sys is populated though, and echo 128 > /sys/class/backlight/amdgpu_bl0/brightness works (also without sudo). Removing the firmware package restored the previous situation (I was able to use X again but not adjust the backlight).

Board footer

Forum Software