Ron / BCLUG via talk said on Wed, 23 Oct 2024 13:02:54 -0700

CAREY SCHUG via talk wrote on 2024-10-23 12:14:

...

winblows

...

wimblows [sic]

Are "winblows" and "M$" still things we do in the 2020s?

Winblows is an OS, M$ is an unpunished former monopolist. SteveT Steve Litt http://444domains.com

a partition table is just binary ones and zeros. there are different kinds of partition tables. it has to read the ones and zeros to determine if, or which kind of, partition table is there. winblows reads those ones and zeros, determines it is NOT a partition table, but that it *IS* a filesystem. why is gparted too unwilling to do that? just as a safety mechanism. If I try to do something, is asks me several times, "are you sure?", all data will be wiped out. maybe if it can, it should warn me there is data on the disk that that will be wiped out if I make any changes. maybe I'm confused on details. is there a function in gparted to figure out what is on the disk? Maybe I need to invoke such an option. or is it a different program? long ago I had a disk with a damaged somehow damaged, and some program tried to recreate filesystems or partitions or something, i don't remember at this point. <pre>--Carey</pre>

On 10/23/2024 3:29 PM CDT D. Hugh Redelmeier via talk <talk@gtalug.org> wrote:

...

From: CAREY SCHUG via talk <talk@gtalug.org>

...

So winblows allows usb (and probably internal) disks to be formated EXFAT but without any partition table.

WHAT POSSIBLE EXCUSE is there for gparted to not recognize this? Obviously winblows can detect there is data, directories, etc on the device, why not GParted?

gparted is a partition editor. It doesn't recognize data, directories, etc. It recognizes partitions and filesystem contained inside partitions.

A disk containing only a filesystem has no partitions.

Since there are no partitions, gparted has nothing to operate on.

What would you have it do?

Perhaps partition the drive? It can do that. But you will lose any existing filesystem.

From what you say elsewhere, this particular filesystem contains nothing of interest yet. So go ahead: partition the drive. Put new, empty filesystem(s) in the new partition(s). --- Post to this mailing list talk@gtalug.org Unsubscribe from this mailing list https://gtalug.org/mailman/listinfo/talk

Your response is helpful Lennart, and this ignorant ex-mainframer even believes he understands. Thank you. before going further, since my disks are empty, based upon this from the internet: MBR is compatible with legacy and older operating systems that do not support GPT. On the other hand, if you have a newer computer with UEFI firmware, GPT is recommended for better compatibility and support for modern features.Jul 11, 2023 should i format with a GPT partition table? IIRC every system I have seen (other than Sun) has had a MBR partition table. to make sure my older computers can access it (hopefuly won't need to), will any operating system that supports 4 terabyte filesystems support GPT? ---------------------------------- practically and realistically, I would still suggest that gparted could and should recognize a few of the most common file-systems, including microsoft and apple. ditto linux itself. As I understand it, linux for some time has recognized GPT partition tables, even though it would always create MBR, so if it doesn't have to recognize microsoft formats, the same argument could be made that it should (historically have) ONLY recognized MBR partition tables. Will linux access an amiga filesystem? Willow? Netware? CP/M? I suspect the number of filesystems a running LINUX can actually access is small enough that adding code to recognize them without a partition table entry is reasonable. Has microsoft changed? Last I knew (thought I knew), it could not access or be aware of ext4 files at all unless special modifications were applied to it. And "irregardless" (intentionally trying to add some levity to this) I would hope WE could be one (or two, or three) better then Microsoft. In practical terms, I am not going to walk into Microcenter and buy a spinning/solid state/usb drive with an Amiga file-system on it. A new and easy to acquire drive will contain --a partition table that identifies the file-systems on it) --an empty partition table --a microsoft filesystem --if removed from a computer sold with linux, an EXT4 or maybe one of a short list, but I suspect ONLY with a partition table. And once I have defined partitions, gparted and/or the linux install process WILL format it at least in linux and (some?) microsoft formats, so it DOES understand them. I have never tried formatting a partition for amiga with gparted... but if so, might that be useful for running an amiga emulator? Since apparently disks with one of a few microsoft filesystems are "common", I think it is reasonable for both Linux AND gparted to recognize them. If I have pulled a disk out of some other computer, I should be expected to have the knowledge of how to check for a filesystem in THAT format. another aside, just out of curiosity: If I find a mainframe storage array (in a dumpster, or "fell off a truck") made from commodity disks, do they have one of the two partition tables, and a filesystem we might know about, or just some hidden binary that only the mainframe controller knows about? <pre>--Carey</pre>

On 10/24/2024 8:13 AM CDT Lennart Sorensen <lsorense@csclub.uwaterloo.ca> wrote:

On Wed, Oct 23, 2024 at 03:51:24PM -0500, CAREY SCHUG via talk wrote:

...

a partition table is just binary ones and zeros.

there are different kinds of partition tables.

it has to read the ones and zeros to determine if, or which kind of, partition table is there.

winblows reads those ones and zeros, determines it is NOT a partition table, but that it *IS* a filesystem.

If you put ext4 on a disk without partitions, windows would not recognize it had a filesystem. Windows can only do that because it recognized the filesystem type and knew what to do with it. Try connecting a drive with an Amiga FFS and see what windows does. Either nothing, or helpfully offers to format the drive for you to make it "usable".

There are hundreds of possible filesystems you could put on there. Why (and even how) could parted know how to recignize them all?

About the best it could do if you go to partition the drive is say 'hey the sector for the partition table is not all zero, are you sure you want to create a new partition table here?'

...

why is gparted too unwilling to do that?

Because it is an impossible task. There is no magic indicator that something is a filesystem. You need to check for the specific header layout of each particular filesystem to detect it. This is why the file tool has a database of magic signatures to recognize all sorts of things. Perhaps parted could instead ask file what it things is on the disk and display that.

For example: lsorense@mythtv64:~$ sudo file -s /dev/sdh /dev/sdh: DOS/MBR boot sector lsorense@mythtv64:~$ sudo file -s /dev/sdg /dev/sdg: DOS/MBR boot sector; partition 1 : ID=0xee, start-CHS (0x0,0,2), end-CHS (0x3ff,255,63), startsector 1, 4294967295 sectors, extended partition table (last) lsorense@mythtv64:~$ sudo file -s /dev/sdg1 /dev/sdg1: Linux Software RAID version 1.2 (1) UUID=e0555ca7:7af8ef1c:d29d1583: e18e5a7 name=mythtv64:1 level=6 disks=9 lsorense@mythtv64:~$ sudo file -s /dev/sda1 /dev/sda1: Linux Software RAID version 1.2 (1) UUID=e0555ca7:7af8ef1c:d29d1583: e18e5a7 name=mythtv64:1 level=6 disks=9 lsorense@mythtv64:~$ sudo file -s /dev/sdf1 /dev/sdf1: Linux Software RAID version 1.2 (1) UUID=e0555ca7:7af8ef1c:d29d1583: e18e5a7 name=mythtv64:1 level=6 disks=9 lsorense@mythtv64:~$ sudo file -s /dev/sdh1 /dev/sdh1: Linux rev 1.0 ext4 filesystem data, UUID=85a0326e-99b6-4a83-9931-6a59cf36140f (needs journal recovery) (extents) (large files) (huge files)

Took me a few tries to find one that wasn't a raid volume. :)

...

just as a safety mechanism. If I try to do something, is asks me several times, "are you sure?", all data will be wiped out. maybe if it can, it should warn me there is data on the disk that that will be wiped out if I make any changes.

maybe I'm confused on details.

is there a function in gparted to figure out what is on the disk?

I don't think so. It's just is just to display, create and modify various partition table formats. Filesystems and data are outside it's scope. It tries to follow the unix philosophy of doing one thing well.

...

Maybe I need to invoke such an option. or is it a different program?

As you see, file is good at telling what things are.

-- Len Sorensen

From: CAREY SCHUG via talk <talk@gtalug.org>

before going further, since my disks are empty, based upon this from the internet:

MBR is compatible with legacy and older operating systems that do not support GPT. On the other hand, if you have a newer computer with UEFI firmware, GPT is recommended for better compatibility and support for modern features.Jul 11, 2023

should i format with a GPT partition table? IIRC every system I have seen (other than Sun) has had a MBR partition table. to make sure my older computers can access it (hopefuly won't need to), will any operating system that supports 4 terabyte filesystems support GPT?

There are lots of partitioning methods. For example *BSD. MBR is old. It was a simple hack based on how the IBM PC booted. Don't use it. Aracana: the primary partition table is embedded in the boot block, the first block of the disk. Any computer that requires MBR is old and obsolete. If you are running a recent Linux on such old hardware, you can still use GPT: Linux will fake up enough of an MBR that a BIOS system will be able to boot from a GPT disk. So: use MBR unless you have some reason not to.

practically and realistically, I would still suggest that gparted could and should recognize a few of the most common file-systems, including microsoft and apple.

You could propose that the the gparted or libparted teams. They might even accept a patch to do it. I personally would not find the idea appealing. A better fix might be to issue a diagnostic that clearly told the user what's up. What it says now didn't work for you. It's not as if gparted / libparted / fatresize is perfect. I find this quite inconvenient: <https://github.com/ya-mouse/fatresize/issues/25>

ditto linux itself.

As has been explained earlier, linux can be configured to do what you want. You could petition your distro to do that. Such a petition could come in the form of a proposed patch.

As I understand it, linux for some time has recognized GPT partition tables, even though it would always create MBR, so if it doesn't have to recognize microsoft formats, the same argument could be made that it should (historically have) ONLY recognized MBR partition tables.

??? Linux doesn't "always create MBR" and hasn't for at least as long as I have cared.

Will linux access an amiga filesystem? Willow? Netware? CP/M? I suspect the number of filesystems a running LINUX can actually access is small enough that adding code to recognize them without a partition table entry is reasonable.

"Linux" isn't one thing. Minimally, it is the kernel. Maximally it is some or all distros. In a random system, chances are that a recognition of a CP/M filesystem is a false positive. Don't do that! Almost NO disks with systems on them are unpartitioned. Only the odd USB stick, flash memory card, and floppy disks (remember those?).

Has microsoft changed? Last I knew (thought I knew), it could not access or be aware of ext4 files at all unless special modifications were applied to it.

So?

And "irregardless" (intentionally trying to add some levity to this) I would hope WE could be one (or two, or three) better then Microsoft.

It depends on what "better" is.

In practical terms, I am not going to walk into Microcenter and buy a spinning/solid state/usb drive with an Amiga file-system on it. A new and easy to acquire drive will contain --a partition table that identifies the file-systems on it) --an empty partition table --a microsoft filesystem --if removed from a computer sold with linux, an EXT4 or maybe one of a short list, but I suspect ONLY with a partition table.

A disk with no partition table is just odd or new. A USB stick or SD card from a store contains stuff that you probably should not trust. Just wipe it.

And once I have defined partitions, gparted and/or the linux install process WILL format it at least in linux and (some?) microsoft formats, so it DOES understand them. I have never tried formatting a partition for amiga with gparted... but if so, might that be useful for running an amiga emulator?

Sadly, Amiga etc. issues are not mainstream.

Since apparently disks with one of a few microsoft filesystems are "common", I think it is reasonable for both Linux AND gparted to recognize them.

As explained, it can be configured to do so. You have been invited to do that configuration.

another aside, just out of curiosity: If I find a mainframe storage array (in a dumpster, or "fell off a truck") made from commodity disks, do they have one of the two partition tables, and a filesystem we might know about, or just some hidden binary that only the mainframe controller knows about?

The word "mainframe" isn't clear. The storage arrays that get thrown out probably don't have SATA drives. They probably have some RAID structure so no single disk has a whole filesystem. The filesystem might well be a Linux one since most SANs are built on Linux. I kind of remember IBM OS/360 disk drive formats: VTOC. Horrible, even 50 years ago. I certainly own disks orphaned by lack of hardware support. ST506 anyone? Anyway, the real problem is probably that we haven't explained how you should think about your problem. With the correct models in you head, this stuff should be easier to understand. Consider keeping notes about your learning venture and write it up. People who know this stuff forget what it was like to learn it. For example, I don't think that there is a signature that lets you reliably recognize a CP/M filesystem (actually, the filesystem evolved). The partition table entry for a partition specifies a partition type. If you don't have a partition table, there is nothing that tells the system that type (that may or may not matter). Summary: if you don't like something in Linux, and the maintainers (of what?) don't seem to care, fix it yourself. In the process, you will probably learn why things are done the way they are. But there is a real chance that you can get your improvement adopted. Try that with Microsoft. Example: <https://github.com/paolostivanin/OTPClient/issues/384>

From: Nick Accad via talk <talk@gtalug.org>

I'm glad you are not top-posting. But you should trim the quote to only contain things you are responding to.

Just to add another point, if you are using LVM, it is normal to use the whole physical disk as a PV without any partitions, and then create the partition table (or not) on the LV.

OK. Another case that gparted should not deal with.

I have personally used many instances where I used the whole disk without any partitions for the file system, it is not as uncommon as you think.

More cases that gparted should not deal with.

Also as someone who works for the largest array maker in the world, I assure you, most arrays do NOT use Linux, unless you are talking about "SAN" as in the switch, in which case, also no, they use proprietary operating systems designed for switches, Linux is relatively new in that area.

Interesting. I know that *BSD is also used. In the early days, Solaris or SunOS was used too, I think. I was explicitly talking about SANs but oblivious to the anatomy. The "switching plane" and the "data plane" are likely different. I admit that I don't know much about them.

On Sat, 26 Oct 2024 10:08:21 -0500 (CDT) CAREY SCHUG via talk <talk@gtalug.org> wrote:

This is a competitive world.

When new devices come out, Linux adds support for them as it can, they don't say "sorry, you'll have to use the competitive operating system for that", granted some require proprietary drivers, so it is not always possible, but generally we (the market) pressure the manufacturer to support Linux.

Carey, I have not played with gparted. I use fdisk to partition drives. More often, I partition drives using the install program of whatever GNU/Linux I am installing. Back in the day, I had an LS-120 floppy drive. The LS-120 disks generally came formatted for Macs, and GNU/Linux had no problems reading them. I have just purchased a new USB stick. It took some fiddling to make the thing mount, and I quickly reformatted it as VFAT, with a label I regard as appropriate. Now everything works perfectly. Could the original format have been XFAT? I don't even know what XFAT is. In the past, I have partitioned my USB drives. I did not bother this time. As I type this, it is mounted on my desktop as /dev/sdi, as opposed to /dev/sdi1. VFAT definitely works in Microsoft Windows. -- Howard Gibson hgibson@eol.ca http://home.eol.ca/~hgibson

I don't think "Linux" is to blame here. Linux is only a kernel, which many distributions (e.g. Debian, Gentoo, etc.) build upon to create a whole operating system. Some distributions cater towards geeks, while others are built with less savvy users in mind. It sounds like you may need a more user-friendly distribution like Ubuntu or Linux Mint. My current distro, Debian XFCE, is less friendly, but even so it includes the Thunar file manager which can automatically mount removable devices. I just tested USB drives both with and without partition tables, and they both mounted just fine. Gparted is a partition editor, so I would not expect it to do more than create and edit partition tables. The "Unix philosophy" that most Linux software follows is to do one job and do it well; you simply need to install (or choose a distro that includes) software for the functionality you desire. A separate issue may be ensuring that new users start with an appropriately friendly distribution... Regards, Clyne On 2024-10-26 11:08, CAREY SCHUG via talk wrote:

This is a competitive world.

When new devices come out, Linux adds support for them as it can, they don't say "sorry, you'll have to use the competitive operating system for that", granted some require proprietary drivers, so it is not always possible, but generally we (the market) pressure the manufacturer to support Linux.

If I can walk into a mass market store and buy a storage device formatted in a way that a major competitive operating system will plug n play, then if and 'as soon as' possible, Linux should also.

Apparently MANY USB storage devices are formatted XFAT with no partition table. If there are other storage devices sold in the after market, formatted such that some other widely used operating system will plug-n-play, then Linux should transparently support them also. Knowing some magic commands to type in as superuser, after first figuring out WHICH usb port the device is on, does not count.

Or are we admitting that Linux is just a niche toy for geeks?

Specifically, Linux the operating system should recognize an XFAT disk without a partition table and use it. What if somebody bought one, loaded a bunch of data onto it from that other OS, then wanted to access it on Linux? Not just Linux sysadmins should be able to use it. Joe non technical bird watcher who buys a linux computer at Microcenter should be able to just plug it in and use it.

And if the Linux OS itself can plug-n-play, then gparted should recognize it too.

Since linux now can resize XFAT partitions, hopefully gparted could also shoehorn a partition table under the filesystem on a storage device that has no partition table. I don't know the details, so maybe this would be too hard.

If a multi petabyte mainframe SAN has a USB connector that can be plugged into MS windows, Apple IOS, Android, or any other mass market computing device and "just work", then Linux (and GParted) should provide the same support.

If there is some format that is ONLY available DIRECTLY from the vendor of a competitive operating system, to people who own and operate that operating system, then this is not a requirement, but would still be nice. For instance, if Oracle sold disks formatted in a way that normally only Sun computers used, but Apple added support in IOS, then Linux should add support too.

--Carey

Steve Litt via talk wrote on 2024-10-27 12:42:

...

Or are we admitting that Linux is just a niche toy for geeks?

Boy, don't I wish Linux would still be just a niche toy for geeks, rather than an increasingly entangled mess aimed at those who refuse to learn.

I think you have that backwards. Linux has grown out of a "niche toy for geeks" into a powerful and dominant platform, and the chorus of complaining is from "those who refuse to learn". They won't learn about why things changed, they won't learn the new ways of doing things, they just want things to stay like they were in nineteen-dippity-doo. Like digital Amish, thinking "technology should've stopped progress at $arbitrary_date" (unspoken: "when I was in my prime and these things excited me. I'm scared of change now").

However, rather than "just a niche toy for geeks", I'd phrase it as "a powerful tool for those willing and able to use that power intelligently."

That's what it's grown into, yes.

Systemd, Gnome, dbus, networkmanager, pulseaudio, and a whole lot more complexifications have been added to Linux in the name of those without the desire or ability to boss their computer instead of having their computer boss them.

None of those technologies cause me any problems, and some make life much better. i.e. I like having multiple audio sources playing at once (video conference and notification, for example). And Wifi connecting seamlessly everywhere.

This complixification has made it hard for people needing to configure/program/script their computer to do things their way.

"Their way" has been deprecated and they are "those who refuse to learn".

The sad thing is this: We added all these ubercomplexifications to attract windows desktop users

Nonsense. Utter nonsense. These "complexifications" were added to solve real world problems. Just because someone with a few static pages on a web site doesn't encounter them does not mean they're entitled to freeze progress. Also, it's odd that complexities are such a frequent complaint topic in forums dedicated to ... *COMPUTERS*. If the complainants were ~10 years older, they'd be railing against computers, "Slide rules are perfectly cromulent and keep working when the power goes out. Don't need to buy a computer, an OS, learn a bunch of apps, plug it in,..."

that goal failed, but now we've weakened Linux for those who need simple POSIX access.

That was *not* the goal. Linux is stronger than it's ever been. Only dogmatism requires POSIX compliance as an end goal. How's BSD, they're POSIX compliant, no? They're at single digits in usage, maybe? Also, Windows now has WSL, supporting many Linux distros inside Windows. That's how thoroughly Linux-with-complexities has won.

Fortunately for me, Void Linux still exists, as does OpenBSD, and if worst comes to worst, Slackware. Sure, use what you prefer, but don't expect anyone who's got real computer problems to solve to care about near religious fervour about some obscure talking points about init systems or something.

Lennart Sorensen said on Sun, 27 Oct 2024 23:18:41 -0400

On Sun, Oct 27, 2024 at 03:42:52PM -0400, Steve Litt via talk wrote:

...

Systemd, Gnome, dbus, networkmanager, pulseaudio, and a whole lot more complexifications have been added to Linux in the name of those without the desire or ability to boss their computer instead of having their computer boss them. This complixification has made it hard for people needing to configure/program/script their computer to do things their way.

Well systemd has solved actual problems. It does a much better job starting and monitoring services than anything else linux has ever had by far.

*For your use case*, not for some mythical other user, what does systemd give you, besides compatibility with software deliberately doctored to use only systemd, than S6 or Runit? What can S6 or Runit not do, that's actually of value to you?

I absolutely don't want to consider going back to any of the garbage init systems we had before.

The preceding statement encompasses a whole lot of init systems. Runit, s6, busybox init, Perp, OpenRC, Epoch, Upstart and sysvinit. Are you saying you've tried every one of them and found each wanting?

Gnome is a disaster these days.

Yep.

network manager is helpful in some cases and hugely annoying in others. Depends what purpose the system serves.

Network Manager's only legitimate benefit, besides being designed for the windows user refusing to learn Linux, is that you can manage your network without root. But other Linux software can give that benefit too.

Pulseaudio always seemed like a confusing buggy mess.

I call Pulseaudio the land of a thousand mutes. [snip]

I have tried the various BSDs. Unfortunately if you have ever used a GNU user space, you can never tolerate the pathetic lack of features in a BSD user space.

Wait a minute. Don't all the GNU utilities and software run on all the BSDs too? All the computer languages work on all BSDs, as does the marvelous ksh shellscripting interpreter. So most of the BSD annoyances you allude to in the preceding paragraph are one tiny shellscript away from solution. SteveT Steve Litt http://444domains.com

Lennart Sorensen said on Thu, 31 Oct 2024 11:09:01 -0400

On Wed, Oct 30, 2024 at 08:37:57PM -0400, Steve Litt via talk wrote:

...

*For your use case*, not for some mythical other user, what does systemd give you, besides compatibility with software deliberately doctored to use only systemd, than S6 or Runit? What can S6 or Runit not do, that's actually of value to you?

S6 I had never even heard of. What system uses that and since when?

If you mean which distros come with s6 as the default init, that would be Adélie Linux (https://www.adelielinux.org/), for the last 5 years. If you mean which on which distros can you install and use S6, that would be any distro that inits with anything other than systemd, or has the option to do so. By the way, the Void Linux distro has shipped with runit as the default init system for at least a decade. One more thing needs to be said. The fact that I consider systemd junk doesn't account for my disdain for it. My disdain is that systemd breaks lots of good things that used to work, adds a ton of complexity, adds a whole lot of attack surface, and integrates itself so far into the OS and user software that it's extremely difficult to remove and replace. I'm a Vim guy but I don't go around bashing Emacs because Emacs does nothing to impede my Vim usage. Please remember this paragraph when you read the rest of my responses.

...

The preceding statement encompasses a whole lot of init systems. Runit, s6, busybox init, Perp, OpenRC, Epoch, Upstart and sysvinit. Are you saying you've tried every one of them and found each wanting?

Yes I have tried some of those and absolutely they all sucked.

^^^^ ^^^ OK, all the ones you tried sucked. I know you never tried s6 because you never heard of it, and I have a feeling you never tried runit either. Those are the two best init systems.

Process monitoring and restarting is absolutely a desirable thing to have and most of them don't do that or at least didn't do it well at the time I tried. Most use shell scripts for things and that's just a crappy system to still use.

Why are shellscripts crappy? They're a wonderful way to boss your computer. Use the ksh scripting language and the shellcheck shellscript validation checker and you'll have a spectacular control over whatever you want. If you consider shellscripts bad, why use a Unix clone at all?

And in the case of software that actually is updated with explicit systemd support you get service monitoring of whether the service is actually ready to work not just whether the process is running. Sometimes it matters that the database is ready to serve not just that it is starting up before you start something else that uses it.

What you have in systemd is a notification of when the process *says* it's ready to work, which may or may not be true information. The way s6 and runit work, you devise an easy test to *prove* the process is ready to work, and act accordingly. Runit has complete monitoring and control of processes via its sv program, and s6 has something equivalent.

I have not seen anything in the other init replacements that does that.

S6 and runit do that. SteveT Steve Litt http://444domains.com

Even better than an editor war: a /sbin/init war! To boil down a long thread: 1) systemd is the default, and integrates a lot of features, but consequently has a large attack surface; 2) there are more minimal init systems (like https://smarden.org/runit/ - pretty interesting) that may be easier to audit and have comparable functionality. I'm not sure I'm willing to commit the required time/focus to switch out systemd with runit, but I'm glad to know there are good options. Thanks, all, for the discussion, ../Dave

Lennart Sorensen via talk said on Fri, 1 Nov 2024 12:47:56 -0400

On Fri, Nov 01, 2024 at 03:47:11AM -0700, Ron / BCLUG via talk wrote:

...

Never, ever heard of it, therefore do not care one iota about it.

I had not either. It says it uses musl libc instead of glibc,

Not uses, *can use* SteveT Steve Litt http://444domains.com

Lennart Sorensen said on Fri, 1 Nov 2024 21:59:49 -0400

On Fri, Nov 01, 2024 at 03:55:56PM -0400, Steve Litt via talk wrote:

...

Not uses, *can use*

The webpage for it explicitly says in the FAQ:

"Adélie uses the musl standard C library."

I took their word for it.

Ohhhhh, I thought you meant Void Linux. I have no idea what standard library Adélie uses, so if they say MUSL, then MUSL it is. Sorry for the misunderstanding. SteveT Steve Litt http://444domains.com

On Tue, Nov 5, 2024 at 7:49 AM Evan Leibovitch via talk <talk@gtalug.org> wrote:

Hi all.

I use my computer to do things -- communicate, manipulate content in various media and formats, organize my life, perform the various tasks required of me. I get involved in "how this is all done" to the minimum extent possible, though I understand that to some exploring the innards of the system is a source of passion for computing as much as it may be to others for cars, coffee, washing machines or the human body.

For most things, I tend to be comfortable with defaults until they stop working or unless I have a specific need for better performance. I try to be sanguine about technology rollovers -- Unix to Linux, X to Wayland, init to systemd, pulseaudio to pipewire, etc etc -- so long as they don't break my ability to do what I need to.

In this context, I am truly intrigued by

...

"Debian's 2014 decision to use systemd without requiring compatibility with any other init system was one of the most sleazy, kangaroo court railroad jobs there's been outside of politics and 1940's southern USA courtrooms."

for I have a hard time comparing the stakes of the two things being compared. In the realm of open source it's not like the Debian debate was between warring corporate special interests each looking to cash out. Is there reason to believe that the various factions in the systemd debate were NOT acting in good faith on behalf of what they thought was the best way forward? One may argue with the process but I don't think that Debian has really suffered much aftermath over the past decade.

Obviously in this world if you don't like something -- for any reasons, technical or not (see "Godot") -- you can fork or re-invent it. So apparently there are multiple re-inventions of init, of which systemd -- like ksh, vim, cups, ext4 and other defaults -- become most popular among mainstream distributions despite the many alternatives available.

I had never heard of Adéline or Void Linux before now. I was aware of Gentoo but have kept my distance. I will likely not give them another look because my application needs define/refine my OS choices. They appear to offer no benefits to my type of use but offer many potential risks if the apps upon which I depend don't like them.

Can someone make an ELI5 case why I should care about systemd alternatives?

I'm no expert but - - - - what I'm understanding is that the complaint is from a philosophical perspective. Unix - - - tools designed to do one thing and doing it well working together to get what one needed done. (In contrast - - - the M$ world loves complexity that is supposed to be hidden from the user because they are not wupposed to know anything about anything except how to pay someone how to set things up etc etc etc.) So along comes systemd - - - - espoused by a more than rather arrogant young 'person' which will remove this area (init systems) that was not as neat nor worked as well as it could have except he was espousing changing things to something quite foreign to 'unix' style thinking - - - everything was supposed to be included in this sandbox. It could be argued that there was some improvement in some things but overall the working philosophy was not just changed - - - said individual trampled upon it and then with loud roars of glee declared said philosophy to be quite dead. (Not true as now there has started an investigation of 'micro services' which as I understand it is just about a 180 degree change from the everything (AND the kitchen sink) which is systemd. Just to make things more interesting said individual is now directly employed by M$ which, at least in my mind, begs some questions. So - - - if you love complexity and would rather have your computer be a black box where you really shouldn't be looking under the hood - - - well systemd is your beastie - - - now if you delight in the interaction of a multitude of various different parts and having at least some access (some may be beyond one - - - that's ok in my world - - - I don't have to know everything) well then not using systemd is your method. (Its like driving a model T or a modern car - - - one you have to think a bit and the other one want to treat you like an idiot so it will try to do all your thinking for you. Heaven help you in stupid conditions if you're in the latter. Think the modern car with its automatic transmission being driven in 6 to 8" of fresh snow and the temperature is just under freezing - - - if you have to stop for some reason - - - good luck getting going again - - - in a manual transmission - - - piece of cake!!!) HTH

On Tue, Nov 5, 2024 at 2:01 PM Evan Leibovitch via talk <talk@gtalug.org> wrote:

On Tue, Nov 5, 2024 at 12:53 PM o1bigtenor via talk <talk@gtalug.org> wrote:

...

[...] the M$ world loves complexity that is supposed to be hidden from the user because they are not wupposed to know anything about anything except how to pay someone how to set things up etc etc etc.

In a former life I made decent money from "people paying someone how to set things up" ... on Linux.

And I've been paid to make things work that the bean counters morphed into cheap trash - - - got to think that such thinking was really really short sighted!

There are people who just don't care about what's under the hood so long as it works. Apple has arguably perfected products for such consumers even better than Microsoft.

Yup and they really enjoy locking you into their changes - - whenever and however they happen. IMO for the multilingual the Mac (early Mac environment - - - haven't been on since before 2000!) was the slickest writing environment I have ever used!

In some ways, then, are we witnessing a mutation of the old Stallman complaint? - Some people prefer free software over proprietary because of its philosophy - Some people prefer open source over proprietary because it just works better and more transparently

Possibly.

If it's open source, it's not hidden. Bob Young's old analogy is recalled with fondness. Just because your car maker won't weld its hood shut, doesn't mean you need to be a mechanic to use it. (paraphrased)

You mean - - -you know how to change the code on the umpteen different control systems - - - - wow I've got this problem w my little p/u . . . .

...

So along comes systemd - - - - espoused by a more than rather arrogant young 'person' which will remove this area (init systems) that was not as neat nor worked as well as it could have except he was espousing changing things to something quite foreign to 'unix' style thinking

A very persuasive person indeed, to swing the preponderance of an entire community behind his path. Or, based on what I can read in the Wiki places, swing Red Hat which put it in Fedora in 2011. It didn't come to Debian until 2014. Then Ubuntu adopted it and I guess you could call it mainstream by then.

Well the company he was then working for likely helped place him in such a fashion that he was 'the guru' - - - IMO Red Hat saw that the change would enhance their market due to tie in and and and so he was given lots of airtime +.

I'm amused that this is still an annoyance to people a decade later.

Well - - - I find that when a 20 something year old that's loud obnoxious and abrasive starts running down people with 25 and 30 years of experience - - - doesn't matter the field the annoyance factor in the group is usually NOT small. No different in this area.

...

said individual trampled upon it and then with loud roars of glee declared said philosophy to be quite dead.

And yet ... after a full-throated debate, the Debian tech committee went with systemd even though that decision caused some developers to resign. Perhaps the functionality trumped the philosophy?

Not really - - - - as I'm remembering it there was a massive push for corporate acceptance from slightly before this. Corporate interests espoused the changes. By the time Debian was voting on it - - it seemed quite a bit like monkey see monkey do - - there was an extreme desire to please corporate think. What's fascinating to me is following what the early Unix developers did - - - yes they were working for major corporate interests - - - but - - they were given very large amounts of autonomy and primarily driven by results. This is something that really hasn't been prevalent since well before even RedHat moving to systemd in 2011. The mantra now is - - - - make money (make money make money chant this for about another 5 minutes) not so much about creating a product that reliably does anything.

After all, there are many other complex systems in a typical Linux distro. Some people still have problems making the audio work well. And let's not even start about desktop environments.

...

Just to make things more interesting said individual is now directly employed by M$ which, at least in my mind, begs some questions.

The only question begged from me is ... in 2024, is the use of "M$" still a thing?

Hmmmmm - - - you find M$ to be benign and caring - - - - if so I suggest you try removing either your painted over glasses or try a lesser dose of the mind bending substances for a bit.

What next? I₿M?

...

(Its like driving a model T or a modern car - - - one you have to think a bit and the other one want to treat you like an idiot so it will try to do all your thinking for you.

These days I hesitate to call people "idiots" for just wanting their tools to work without needing to know how. Perhaps their expertise and/or enthusiasm lies elsewhere. Indeed maybe this seems also a debate between those who see Linux as just a tool to do something else and those who care about the OS as an end in itself.

Well - - - we've gotten to where one is supposed to be helpless and dependent on experts for everything. Guess that I like doing, love learning and have found far too many experts who really only know how to wield a large shovel and really don't know what they're talking about.

...

Heaven help you in stupid conditions if you're in the latter. Think the modern car with its automatic transmission being driven in 6 to 8" of fresh snow and the temperature is just under freezing - - - if you have to stop for some reason - - - good luck getting going again - - - in a manual transmission - - - piece of cake!!!)

Maybe there's a good parallel here. The simple method has fallen out of favour (I think latest figures have manuals at 2% of sales in North America), while an even more-complex solution (all-wheel and four-wheel drive) has become dominant. It addresses the edge case above without subjecting the user to the hell of using a manual transmission in bumper-to-bumper urban traffic on a hill (a far more common situation for many).

Except by that thinking all those vehicles should be controlled by a central computer to make sure everything works well. The standing joke is that the vast majority of all-wheel and 4 wheel drive vehicles never ever see off road. I have had 1 - 4 wheel drive - - - learned more bad habits with that thing than much else and also the inadequacies of such. Have found its much much more about knowing what you're doing and far less about having perfect tools.

I appreciate the high-level explanation of this as being one of simplicity versus complex innovation, but there was a lot more here.

Oh yes there was - - - I am surprised that you don't remember any of this - - - I'm quite sure you were around through all of this time. I was trying to keep things at least somewhat polite and trying not to just rant. (Some of the comments at the time from some individuals were really not the kind of thing that should have even been thought even if the lightning rod was an incredibly brass obnoxious quite young person.) Regards

On Tue, Nov 5, 2024 at 4:21 PM William Park via talk <talk@gtalug.org> wrote:

On 2024-11-05 12:52, o1bigtenor via talk wrote:

...

Think the modern car with its automatic transmission being driven in 6 to 8" of fresh snow and the temperature is just under freezing - - - if you have to stop for some reason - - - good luck getting going again - - - in a manual transmission - - - piece of cake!!!)

Can you explain in more detail? I was thinking about buying manual Kia Rio (cheapest new car on the market, now no longer sold). My automatic has 1, 2, 3, 4, D, N, R, and P. Why is manual different? --

In an automatic transmission - - - who chooses when its time to shift? In conditions like I listed on a 5 speed I would likely start in 3rd - - - its tricky but you don't want the tires to spin - - - - at all. Any spinning and you make ice and you're done - - - time for shoveling and if you've got a heavy snowfall - - - -well you may also be pushing snow across the undercarriage (a LOT more shoveling!!). If you've never driven conditions like this you likely won't find the idea possible. (Your automatic transmission even in 3rd will start you off in 1st - - - that is the programmed function - - - maybe it won't but then that's an unusual automatic transmission!) Regards

On Wed, Nov 06, 2024 at 05:51:21AM -0600, o1bigtenor via talk wrote:

In an automatic transmission - - - who chooses when its time to shift?

In conditions like I listed on a 5 speed I would likely start in 3rd - - - its tricky but you don't want the tires to spin - - - - at all. Any spinning and you make ice and you're done - - - time for shoveling and if you've got a heavy snowfall - - - -well you may also be pushing snow across the undercarriage (a LOT more shoveling!!). If you've never driven conditions like this you likely won't find the idea possible.

(Your automatic transmission even in 3rd will start you off in 1st - - - that is the programmed function - - - maybe it won't but then that's an unusual automatic transmission!)

With the power of car engines these days, many can easily spin their tires in 3rd gear. Starting in a higher gear just means you now have the car wanting to go much faster at idle than you are going and hence you likely will spin the tires. Stuff that made sense 30 or 40 years ago in cars does not always apply today. And we usually have traction control to help out these days too. In my first car which was a manual, engine idling in 1st gear was about 10km/h. Going less than that meant slipping the clutch. In my current (and previous) car doing 1km/h is trivial, because it's a hybrid and electric motors work just fine at low speeds. Much easier to start slowly or go up steep parking garage ramps with that. Winter driving is a heck of a lot easier too. -- Len Sorensen

On Tue, Nov 5, 2024 at 9:53 AM o1bigtenor via talk <talk@gtalug.org> wrote:

On Tue, Nov 5, 2024 at 7:49 AM Evan Leibovitch via talk <talk@gtalug.org> wrote:

...

Hi all.

I use my computer to do things -- communicate, manipulate content in various media and formats, organize my life, perform the various tasks required of me. I get involved in "how this is all done" to the minimum extent possible, though I understand that to some exploring the innards of the system is a source of passion for computing as much as it may be to others for cars, coffee, washing machines or the human body.

For most things, I tend to be comfortable with defaults until they stop working or unless I have a specific need for better performance. I try to be sanguine about technology rollovers -- Unix to Linux, X to Wayland, init to systemd, pulseaudio to pipewire, etc etc -- so long as they don't break my ability to do what I need to.

In this context, I am truly intrigued by

...

"Debian's 2014 decision to use systemd without requiring compatibility with any other init system was one of the most sleazy, kangaroo court railroad jobs there's been outside of politics and 1940's southern USA courtrooms."

for I have a hard time comparing the stakes of the two things being compared. In the realm of open source it's not like the Debian debate was between warring corporate special interests each looking to cash out. Is there reason to believe that the various factions in the systemd debate were NOT acting in good faith on behalf of what they thought was the best way forward? One may argue with the process but I don't think that Debian has really suffered much aftermath over the past decade.

Obviously in this world if you don't like something -- for any reasons, technical or not (see "Godot") -- you can fork or re-invent it. So apparently there are multiple re-inventions of init, of which systemd -- like ksh, vim, cups, ext4 and other defaults -- become most popular among mainstream distributions despite the many alternatives available.

I had never heard of Adéline or Void Linux before now. I was aware of Gentoo but have kept my distance. I will likely not give them another look because my application needs define/refine my OS choices. They appear to offer no benefits to my type of use but offer many potential risks if the apps upon which I depend don't like them.

Can someone make an ELI5 case why I should care about systemd alternatives?

I'm no expert but - - - - what I'm understanding is that the complaint is from a philosophical perspective.

Unix - - - tools designed to do one thing and doing it well working together to get what one needed done.

(In contrast - - - the M$ world loves complexity that is supposed to be hidden from the user because they are not wupposed to know anything about anything except how to pay someone how to set things up etc etc etc.)

So along comes systemd - - - - espoused by a more than rather arrogant young 'person' which will remove this area (init systems) that was not as neat nor worked as well as it could have except he was espousing changing things to something quite foreign to 'unix' style thinking - - - everything was supposed to be included in this sandbox. It could be argued that there was some improvement in some things but overall the working philosophy was not just changed - - - said individual trampled upon it and then with loud roars of glee declared said philosophy to be quite dead. (Not true as now there has started an investigation of 'micro services' which as I understand it is just about a 180 degree change from the everything (AND the kitchen sink) which is systemd. Just to make things more interesting said individual is now directly employed by M$ which, at least in my mind, begs some questions.

I do like your sense of imagination here. Now, I was actually present, when we were designing systemd, for some of the discussions. The only thing we were trying to do was fix a real problem - which users had complained about. No grand plans of how we are going to bring all this complexity and become like Microsoft world. If there was a clear design decision we made, it was systemd was for Linux and would use Linux features and not try to remain "unixy" and just use the bare subset but be cross platform. Now, I am sure Lennart and I will have a great laugh about how he went out to become a dictator with the use of systemd and how our evil plan succeeded. But seriously - let's stop with these complicated conspiracy theories. I was there when we talked about how/why systemd. A fact that is not known too well is Lennart did go with some of these ideas to then init systems and they were not interested in implementing them, so he just scratched his itch. Turns out he was right because systemd has now been adopted and is actually (surprise surprise!) liked by administrators who do something more complicated than a home setup. Dhaval

On Fri, Nov 01, 2024 at 04:36:53AM -0400, Steve Litt via talk wrote:

If you mean which distros come with s6 as the default init, that would be Adélie Linux (https://www.adelielinux.org/), for the last 5 years. If you mean which on which distros can you install and use S6, that would be any distro that inits with anything other than systemd, or has the option to do so.

By the way, the Void Linux distro has shipped with runit as the default init system for at least a decade.

One more thing needs to be said. The fact that I consider systemd junk doesn't account for my disdain for it. My disdain is that systemd breaks lots of good things that used to work, adds a ton of complexity, adds a whole lot of attack surface, and integrates itself so far into the OS and user software that it's extremely difficult to remove and replace. I'm a Vim guy but I don't go around bashing Emacs because Emacs does nothing to impede my Vim usage. Please remember this paragraph when you read the rest of my responses.

Initially I also thought systemd was too complex. Having worked with it for a while I eventually figured out why it had changed a bunch of stuff the way it did. The stuff it replaced was in fact not working well, it was just traditional. As for attack surface, I am not convinced that is true. The amount of stupid mistakes one can make in a shell script that can be exploited is huge. systemd does not prevent people from supplying other startup scripts for daemons. Of course the fact most people seem to now be fairly happy with systemd means there are a lot less people who want to spend time writing such scripts. Blaming systemd for being successful is unfair. I haven't seen any daemon that can't run without systemd, but certainly many do support it now. Certainly in the case of Debian there has been a lot of arguing over systemd, and there are definitely other init systems available, but systemd is the one installed by default. And yes there are packages that no longer seem to provide sysvinit startup scripts because the package maintainer doesn't want to spend time on it since they don't use it and don't even have a system setup to test it on.

OK, all the ones you tried sucked. I know you never tried s6 because you never heard of it, and I have a feeling you never tried runit either. Those are the two best init systems.

It appears from the git tree that it came out in 2014 (version 2.0.0, I can't find any indication of a version before that). That's a bit late to the party as most major distributions had already gone to systemd by then. Even Debian had by that time and they were one of the last to do it.

Why are shellscripts crappy? They're a wonderful way to boss your computer. Use the ksh scripting language and the shellcheck shellscript validation checker and you'll have a spectacular control over whatever you want. If you consider shellscripts bad, why use a Unix clone at all?

There is far far more to a unix system than shell scripts. Even a python script would give better control over things than a shell script. It seems a bit overkill though. Dealing with arguments and quoting and such in a shell script is very error prone. systemd avoids that by not using a shell to launch processes at all. That is a lot safer. Less flexible, but much safer and less error prone. I want my system to work. Building from source and tweaking was fun and interesting at the start, but that was 30 years ago. To me distributions like Gentoo are a joke. Their focus is something that is a complete waste of time. There is no point building everything from source on every machine. There is no benefit just wasted CPU cycles. I can't recall which linux distribution used to claim that it made a difference that you optimized everything for your particular CPU model at compile time, but again, what a waste of time. Looking at the packaging file to see how things are being built is definitely interesting and you can learn from it, but that doesn't mean you need to actually spend time doing it to learn.

What you have in systemd is a notification of when the process *says* it's ready to work, which may or may not be true information. The way s6 and runit work, you devise an easy test to *prove* the process is ready to work, and act accordingly. Runit has complete monitoring and control of processes via its sv program, and s6 has something equivalent.

I would think if it isn't true then that is a bug in the service that should be fixed. And making an external test could be rather complicated. Sounds annoying. A proper test is not a bad idea, but I am not sure the complications of implementing it are worth it.

S6 and runit do that.

Well it seems s6 was too late to be relevant. Runit predates systemd by about 6 years. There must have been some reason it didn't catch on instead of sysvinit at the time. -- Len Sornsen

Steve Litt via talk wrote on 2024-11-01 12:55:

Here's my runit run script for sshd:

============================= #!/bin/sh exec 2>&1 ssh-keygen -A > /dev/null 2>&1 [ -r conf ] && . ./conf exec /usr/bin/sshd -D $OPTS =============================

The ss-keygen line generates host keys only if there are none. There's no conf file so the conf line is a no-op. Because there's no ./conf, $OPTS is an empty string. The -D in sshd runs it in the foreground, which is the right way to do things in an init system. A shebang plus four lines. Yeah, it's possible to screw up, but you have to be pretty careless to screw it up.

What does the `exec 2>&1` on a line after the shebang do? How does that script handle restarts? If sshd goes down, what happens? My system will restart on failure unless exit code 255 from sshd occurred: Restart=on-failure RestartPreventExitStatus=255

I'm blaming it for deliberately becoming inseparable from other software

The blame lies with "the other software", whose developers probably just want to support the most common environment. I can't blame them. Again, when one is receiving something for free, it's unseemly to demand it be given in a manner that increases the burden of the giver.

and through marketing convincing software authors to include systemd-only code

Conspiratorial codswallop. Devs want to dev, not faff about with init systems. It's not like "systemd™ Inc.®" spent advertising dollars targeting devs.

which led to the recent supply chain attack against the recent CVE-2024-3094 exploit that came with hours of wrecking the Internet. Note that only xz and liblzma packages with systemd as a dependency suffered this.

Note also that systemd was in the process of releasing statically linked versions of their binaries to prevent such attacks, which rushed the malware. (Which will lead to retro-techno folks to moan about statically linked binaries.)

Debian's 2014 decision to use systemd without requiring compatibility with any other init system was one of the most sleazy, kangaroo court railroad jobs there's been outside of politics and 1940's southern USA courtrooms.

Hyperbole much?

It's been 10 years. Runit is tried and true. It's superior.

Opinion, not fact established by evidence.

You should check it out.

I am (we are?) satisfied with what we've got, so that seems like a waste of time. As Evan(?) said, we're trying to get stuff done, not bask in the ideological purity of the underlying software. As long as it's FLOSS, don't really care. And, systemd is FLOSS.

...

...

What you have in systemd is a notification of when the process*says* it's ready to work, which may or may not be true information. The way s6 and runit work, you devise an easy test to*prove* the process is ready to work, and act accordingly. Runit has complete monitoring and control of processes via its sv program, and s6 has something equivalent.

I would think if it isn't true then that is a bug in the service that should be fixed.

*Should* be. I'm not holding my breath.

I'm still waiting for a citation that "What you have in systemd is a notification of when the process*says* it's ready to work", because that Is. Not. True. Unless I missed something, was there a reply to the several types of execution "Type=" I posted earlier?

...

And making an external test could be rather complicated.

========================================== if ! ping -c2 8.8.8.8 > /dev/null; then sleep 1 exit 1 fi ==========================================

In the preceding test for a working network, substitute whatever IP address you consider "always up" for 8.8.8.8. I think you can get away without the sleep, but I keep it in there just to make sure I don't create a race condition.

So, if a 2-ping test fails, sleep one second (why, hasn't it already failed?), and exit 1 (a failure state) is the solution? Why the `sleep 1`? What if it takes longer than 2 ping packets for the network to come up for some reason? What if 30 seconds is the desired wait time? This will wait for the network to come up:

After=network.target

and

TimeoutSec=90

will fail the startup after 90 seconds.

Not complicated at all. For instance, if it's a Postgres database, just execute a psql command the performs a quick and simple SELECT statement that proves or disproves that a database is ready to use.

Just one test? What if the DB is in recovery mode and takes a while? Loop the calling of `psql`? How many times, for how long? Such a hack of a method when what we have is so much better.

...

There must have been some reason it didn't catch on instead of sysvinit at the time.

F o l l o w t h e m o n e y .

The systemd takeover was bankrolled by Red Hat, a hugely capitalized giant.

I pay nothing for systemd. Canonical users pay nothing for systemd. No one pays for systemd. RedHat initially went with Canonical's Upstart until systemd was deemed better. If RedHat went with Runit, they'd still sell support.

Summary: Red Hat requires complexity to make money.

Then you'd think Canonical and Suse would have gone with Runit, won over all the pissed off ex-RedHat clients, but that didn't happen. Silly conspiracy is silly.

Seriously, install Void Linux on a VM or unused hardware.

Void Linux is probably fine, but no amount of evangelizing is going to win converts who are just trying to get shit done and don't want to be techno hipsters.

Steve Litt via talk wrote on 2024-11-01 12:55:

Have you seen all the different variations of keys that can go in a unit file? Have you noticed that a lot of them seem to do the same thing on viewing the names

Often times concepts are related but different. This can lead to names that are similar, but different. Especially true in complex fields, which services management is. Claiming otherwise is just hand-waving away a bunch of actual issues.

but you need to read some pretty voluminous documentation

Now having "voluminous documentation" is a bad thing? It's *really* superb documentation. Far better than trying to parse a shell script and remember what each bit of arcane punctuation does.

to understand when to use each key. Back in the day, we used to say, "RTFM" to those unwilling to learn how software works.

I guess now, "Ask ChatGPT". Learning (and using) all the obscure commands and syntax of a shell script (better feed it to an external site so it doesn't blow up in my face... https://www.shellcheck.net/ ) is fine, but RTFingM is a bridge too far?

On Tue, Nov 5, 2024 at 9:21 PM Ron / BCLUG via talk <talk@gtalug.org> wrote: snip

Back in the day, we used to say, "RTFM" to those unwilling to learn how software works.

You bet - - read the documents written by experts for experts so that you can understand what is going on. Besides most of such documents were written by those who whilst being excellent coders were terrible written word communicators.

I guess now, "Ask ChatGPT".

Be a wonderful solution if one could get access to it! (Need to supply a working stupid phone connection to verify identity and, where I live - - - - well I equate the finding of wireless signals with that of finding honest politicians so its a no dice - - - at all but then since I've determined that I really don't want to get mired in mediocrity and that seems to in a nutshell be what AI is about.) Regards

Steve Litt via talk wrote on 2024-10-30 17:37:

*For your use case*, not for some mythical other user, what does systemd give you, besides compatibility with software deliberately doctored to use only systemd, than S6 or Runit? What can S6 or Runit not do, that's actually of value to you?

1) Well supported software that's used by so-called enterprise orgs. Same reason I began using Linux. 2) Something that works well and stays out of my way. 99% of the time it shouldn't even be noticeable. 3) Lots of configuration options.

compatibility with software *deliberately doctored* to use only systemd

"I want to use your free software, and I demand it be provided in the manner of my liking."

...

I absolutely don't want to consider going back to any of the garbage init systems we had before.

The preceding statement encompasses a whole lot of init systems. Runit, s6, busybox init, Perp, OpenRC, Epoch, Upstart and sysvinit. Are you saying you've tried every one of them and found each wanting?

He probably hasn't, I certainly haven't, because I (many / most of us) use our computers to accomplish tasks, not to become an init-system-hipster in service of a dogmatic belief system. Trying out every niche init system seems like a complete waste of time and even if I encountered one that I really liked, then what?

Network Manager's only legitimate benefit, besides being designed for the windows user refusing to learn Linux

This is really dumb. Anything that makes Linux easier to use is making Linux users' lives easier. I mean, what is wrong with any of these features? (from the Arch Wiki):

NetworkManager is a program for providing detection and configuration for systems to automatically connect to networks. NetworkManager's functionality can be useful for both wireless and wired networks. For wireless networks, NetworkManager prefers known wireless networks and has the ability to switch to the most reliable network. NetworkManager-aware applications can switch from online and offline mode. NetworkManager also prefers wired connections over wireless ones, has support for modem connections and certain types of VPN.

Or, from https://networkmanager.dev/ :

NetworkManager is the standard Linux network configuration tool suite. It supports large range of networking setups, from desktop to servers and mobile and integrates well with popular desktop environments and server configuration management tools.

How someone can turn that into "because Windows users refuse to learn Linux" is incomprehensible to me.

From: Steve Litt via talk <talk@gtalug.org>

D. Hugh Redelmeier via talk said on Fri, 25 Oct 2024 11:51:42 -0400

Linux was a simple hack based on Unix.

Not really. Everything starts small, but Linux was quite large by the time it was really useful for something other than Linux development.

Use it for your boot drive if you can. UEFI is a mess even in its specifications, hardware vendors' botched implementation makes it worse, and its attack surface is breathtaking compared to MBR. There's nothing stopping you from using the boot drive MBR and the rest of your drives the very nice GPT (but not UEFI).

As far as I know, every machine that has UEFI implements MBR booting through UEFI. In particular, the "CSM" implements BIOS calls. I don't think that you are avoiding vulnerabilities using MBR on a machine with UEFI. MBR's primary partitions are actually described by a few entries in the last bytes of the boot sector. There is only room for four entries so the additionally hack of "extended partition" was invented. Where does you boot loader live in MBR? The primary boot-loader loads a secondary boot-loader from the partition to be booted. Where does the secondary boot loader (which must fit in a single block!) get the rest of the machinery for booting? Normally from an list of absolute block numbers (very fragile). UEFI has a less hacky answer: the ESP is a full fledged file-system allowing files to be used. These hacks can mean perplexing behaviour unless you understand how it is all put together. Secure Boot, which comes with UEFI, is a little bit of a comfort. Not perfect.

On Wed, Oct 30, 2024 at 06:40:02PM -0400, D. Hugh Redelmeier via talk wrote:

Not really. Everything starts small, but Linux was quite large by the time it was really useful for something other than Linux development.

As far as I know, every machine that has UEFI implements MBR booting through UEFI. In particular, the "CSM" implements BIOS calls.

CSM is gone on newer machines.

I don't think that you are avoiding vulnerabilities using MBR on a machine with UEFI.

MBR's primary partitions are actually described by a few entries in the last bytes of the boot sector. There is only room for four entries so the additionally hack of "extended partition" was invented.

Where does you boot loader live in MBR? The primary boot-loader loads a secondary boot-loader from the partition to be booted. Where does the secondary boot loader (which must fit in a single block!) get the rest of the machinery for booting? Normally from an list of absolute block numbers (very fragile).

UEFI has a less hacky answer: the ESP is a full fledged file-system allowing files to be used.

These hacks can mean perplexing behaviour unless you understand how it is all put together.

Secure Boot, which comes with UEFI, is a little bit of a comfort. Not perfect.

Yeah for most things UEFI is a better design. They did have a lot of years to think about how to fix the mess of the MBR boot process. -- Len Sorensen

On Fri, Oct 25, 2024 at 02:59:48AM -0500, CAREY SCHUG wrote:

Your response is helpful Lennart, and this ignorant ex-mainframer even believes he understands. Thank you.

before going further, since my disks are empty, based upon this from the internet:

MBR is compatible with legacy and older operating systems that do not support GPT. On the other hand, if you have a newer computer with UEFI firmware, GPT is recommended for better compatibility and support for modern features.Jul 11, 2023

should i format with a GPT partition table? IIRC every system I have seen (other than Sun) has had a MBR partition table. to make sure my older computers can access it (hopefuly won't need to), will any operating system that supports 4 terabyte filesystems support GPT?

Well the largest disk MBR can handle (assuming it uses 512 byte sectors which most still do), is 2TB. Anything bigger has to use GPT.

practically and realistically, I would still suggest that gparted could and should recognize a few of the most common file-systems, including microsoft and apple.

ditto linux itself.

As I understand it, linux for some time has recognized GPT partition tables, even though it would always create MBR, so if it doesn't have to recognize microsoft formats, the same argument could be made that it should (historically have) ONLY recognized MBR partition tables.

The linux kernel has support GPT since the late 90s since it was part of EFI for the Itanium so it had to be supported to work with that architecture. Later when x86 moved to UEFI GPT came along.

Will linux access an amiga filesystem? Willow? Netware? CP/M? I suspect the number of filesystems a running LINUX can actually access is small enough that adding code to recognize them without a partition table entry is reasonable.

To a large extent the answer is that yes it can. For example: mythtv64:~/gp# find /lib/modules/6.11.2-amd64/kernel/fs/affs/ /lib/modules/6.11.2-amd64/kernel/fs/affs/ /lib/modules/6.11.2-amd64/kernel/fs/affs/affs.ko.xz mythtv64:~/gp# modinfo affs filename: /lib/modules/6.11.2-amd64/kernel/fs/affs/affs.ko.xz license: GPL description: Amiga filesystem support for Linux I think warning about not blank is enough. You can then go use file to check what it contains and get much better results from a tool designed and maintained for exactly the purpose of identifying things. But certainly you could expect to encounter a drive without partitions with some file system (NTFS, exFAT, FAT or even HFS+) or a partition table, usually MBR or GPT (since the mac has used GPT for years now ever since moving to x86 and not ARM processors). Hitting a partition table from Amiga or BSD or AIX or something else is highly unlikely.

Has microsoft changed? Last I knew (thought I knew), it could not access or be aware of ext4 files at all unless special modifications were applied to it.

Definitely no change there that I am aware of.

And "irregardless" (intentionally trying to add some levity to this) I would hope WE could be one (or two, or three) better then Microsoft.

In practical terms, I am not going to walk into Microcenter and buy a spinning/solid state/usb drive with an Amiga file-system on it. A new and easy to acquire drive will contain --a partition table that identifies the file-systems on it) --an empty partition table --a microsoft filesystem --if removed from a computer sold with linux, an EXT4 or maybe one of a short list, but I suspect ONLY with a partition table.

What if it contains an MBR and a GPT partition table? They are not in the same sector locations by design. What if it has GPT but the two copies are not in sync?

And once I have defined partitions, gparted and/or the linux install process WILL format it at least in linux and (some?) microsoft formats, so it DOES understand them. I have never tried formatting a partition for amiga with gparted... but if so, might that be useful for running an amiga emulator?

Since apparently disks with one of a few microsoft filesystems are "common", I think it is reasonable for both Linux AND gparted to recognize them.

If I have pulled a disk out of some other computer, I should be expected to have the knowledge of how to check for a filesystem in THAT format.

another aside, just out of curiosity: If I find a mainframe storage array (in a dumpster, or "fell off a truck") made from commodity disks, do they have one of the two partition tables, and a filesystem we might know about, or just some hidden binary that only the mainframe controller knows about?

Of course parted is far from the only partitioning tool on linux. It is certainly NOT the one I would pick to use if partitioning a drive. cfdisk is so much more pleasant to work with. I will use gparted if I need to resize partitions on a drive though. About the only think I will consider using it for. Disk partitioning is a somewhat complicated thing, although at least on modern machines it is much much nicer than the insanity you used to have to do on Solaris or IRIX or those other old unix systems where you had to tell it the drive geometry and then create disk labels all manually. -- Len Sorensen

my fat baby finger hits ctl instead of shift and random things happen, including "send" when an email is far from complete. is there a way to grab something back from the list server? for emails I create from scratch, i don't put in the "to" until done, harder to do on a reply is there a settings to turn off some or many ctl-hotkeys in a browser?

On Sun, Oct 27, 2024 at 08:07:04AM -0500, CAREY SCHUG wrote:

Thank you again Lennart, this ignorant mainframer is slowly being dragged kicking and screaming into the modern age.

Did IBM mainframes ever get around to using ascii or are they still ebdbic?

1. Linux (the whole thing, not the kernal) should handle random hardware as well as any major competitiors, at least windows and IOS.

--any commodity devices sold at retail should just work. I

Well given the hardware makers get to write drivers with access to specifications and submit them to Microsoft for automatic install on windows, while Linux developers have to write the drivers themselves, often without access to documentation, I would say that is very much not a fair expectation at all. As for IOS, not sure. Do people regularly connect things to their iphone or ipad? Do they actually work and do anything if they do? Assuming you didn't mean CISCO IOS. Lots of hardware has been made where instead of following the published standard, they just made it work with whatever windows did, which was a subset of the standard or in some cases even potentially in violation of the standard. An OS that actually implemented the standard would then not work with said hardware because it was made wrong. Quite a few drivers in linux have tweaks in them for specific hardware because that hardware didn't correctly follow standards. Someone had to figure that out and implement the workaround. Now for the automounting, certainly some systems will, some won't. https://superuser.com/questions/1775441/usb-stick-without-partition-maybe-bu... says Debian running gnome automounted a USB drive without partitions for them. A beaglebone running Debian but not gnome did not. So it depends on the automounter being used whether it supports recognizing and mounting unpartitioned drives that are inserted. Interestingly it also appears in that post that parted does in fact recognize a drive with just a filesystem. Of course gparted is not parted, it's a gui wrapper around parted and some other utilities. Perhaps gparted just failed to implement support for something parted already did. Or it is a question of the version of parted. Older parted did filesystem stuff, and they decided to remove all that in 3.x releases. Seems someone decided that parted should not do filesystem stuff, only partition stuff. -- Len Sorensen

CAREY SCHUG via talk said on Fri, 25 Oct 2024 02:59:48 -0500 (CDT)

Your response is helpful Lennart, and this ignorant ex-mainframer even believes he understands. Thank you.

before going further, since my disks are empty, based upon this from the internet:

MBR is compatible with legacy and older operating systems that do not support GPT. On the other hand, if you have a newer computer with UEFI firmware, GPT is recommended for better compatibility and support for modern features.Jul 11, 2023

should i format with a GPT partition table? IIRC every system I have seen (other than Sun) has had a MBR partition table. to make sure my older computers can access it (hopefuly won't need to), will any operating system that supports 4 terabyte filesystems support GPT?

I'll give you my opinion. Others vary. There's no way that the OS plus software, config and cache take more than 1TB, so I make my root partition on an NVMe either 1TB or 500GB. Then I bind mount the rest of the mount points with space on a 14TB, 7200 RPM spinning rust. That 14GB is of course formatted GPT, and because it doesn't boot, it needn't have UEFI for the spinning rust. Meanwhile, the small NVMe or SSD doesn't need GPT, so it can be partitioned the old MBR way So if, and this is a big if, your motherboard supports MBR well, in my opinion this is currently the best way to go. However, note that some motherboards make it very difficult to boot from USB or DVD via MBR, sometimes requiring a trip through the BIOS setup screen. Also, many or most bootable DVDs and USBs can boot MBR *or* UEFI, but not both. SteveT Steve Litt http://444domains.com

On Mon, Oct 28, 2024 at 03:19:02AM -0400, D. Hugh Redelmeier via talk wrote:

Why do you consider MBR better than GPT for this small drive?

Good question. I can't think of anything MBR does better than GPT including booting. -- Len Sorensen

D. Hugh Redelmeier said on Mon, 28 Oct 2024 03:19:02 -0400 (EDT)

...

From: Steve Litt via talk <talk@gtalug.org>

...

Meanwhile, the small NVMe or SSD doesn't need GPT, so it can be partitioned the old MBR way So if, and this is a big if, your motherboard supports MBR well, in my opinion this is currently the best way to go.

Why do you consider MBR better than GPT for this small drive?

First, we need to dispense with some unfortunate terminology so we're all on the same page. MBR and GPT are two different *partition styles*. In most situations, GPT is a much better choice than MBR. GPT addresses huge disks, can accommodate up to 128 partitions, with each partition capable of sizing to 256 TB. In most cases, GPT is vastly superior to MBR. Now let's talk about boot modes. Boot mode is what happens very early in the boot, well before the initramfs gets run. The two most popular boot modes in Linux are UEFI and Legacy. Legacy works with MBR disks. UEFI works only with GPT disks. Legacy boot is OK: Could be better, but it's OK. UEFI is a stinking steaming pile of pig produced fertilizer that's almost unfathomable, can permanently brick your computer, is difficult to boot pluggable media with, and makes it difficult for people using their computer in ways Dell, HP, Acer or Asus didn't intend it to be used. Oh yeah, and UEFI is so complicated that the vendors have made mistakes in their UEFI implementations that software can permanently brick the computer, especially when Lennart Poettering made the brilliant decision to let systemd write to the EFI partition. So my opinion of partition schemes is the following: GPT is great, MBR is good. My opinion of boot styles is the following: Legacy is OK, and UEFI is junk. On all drives except the boot drive, boot styles aren't used and don't matter, so on non-boot drives I use GPT because I think it's superior and because I buy drives bigger than 2TB. But on boot drives, I try very hard to go legacy boot style, and the only way to have legacy boot style is for that boot drive to have an MBR partition scheme. UEFI has the following advantages over Legacy boot: * Secure boot: If secure boot is a must, then UEFI is a must, but for me and a lot of others, Secure Boot is a must-not. * UEFI mandated GPT works with huge disks, whereas MBR works only up to 2TB. If you simply must have a larger boot disk, you have to go UEFI. Me, I use a small boot disk with everything but /usr being mount points from a GPT partitioned 14GB spinning rust. * UEFI mandated GPT has more than 8 partitions: This might have been a problem before the invention of bind mounts, but it's not now. You need a root partition, a swap partition (actually now days you can use a swap file), and one partition for each physical disk. Use bind mounts for all the things that used to be partitions, like /home/hugh or /var. It's mighty difficult to use up your 8 partitions. * UEFI is reputed to boot up faster. Your mileage may differ, but on my computer the worst culprit in slowing down boot is the stupid initramfs. Those seeking quicker boot should put their disk format drivers compiled into the kernel, and forego the initramfs. In summary, terminology is important: Partitioning schemes MBR GPT Boot styles Legacy UEFI UEFI is junk and works only with GPT, so when I have the opportunity, I use MBR partitioning scheme with its Legacy boot style for the boot disk, and GPT (with no boot style) for all other disks. HTH, SteveT Steve Litt http://444domains.com

On Sun, Oct 27, 2024 at 11:52:44PM -0400, Steve Litt via talk wrote:

I'll give you my opinion. Others vary.

There's no way that the OS plus software, config and cache take more than 1TB, so I make my root partition on an NVMe either 1TB or 500GB. Then I bind mount the rest of the mount points with space on a 14TB, 7200 RPM spinning rust. That 14GB is of course formatted GPT, and because it doesn't boot, it needn't have UEFI for the spinning rust. Meanwhile, the small NVMe or SSD doesn't need GPT, so it can be partitioned the old MBR way So if, and this is a big if, your motherboard supports MBR well, in my opinion this is currently the best way to go.

However, note that some motherboards make it very difficult to boot from USB or DVD via MBR, sometimes requiring a trip through the BIOS setup screen. Also, many or most bootable DVDs and USBs can boot MBR *or* UEFI, but not both.

As far as I know most machines made in the last 5 or so years do not support anything other than UEFI booting anymore. Servers seem to be still offering it, but as far as I can tell starting either this year or next year, they are going UEFI only as well. At least if they have an intel CPU. -- Len Sorensen