| View previous topic :: View next topic |
| Author |
Message |
krinn
Watchman
Joined: 02 May 2003
Posts: 7470
|
 Posted: Tue Nov 13, 2018 1:23 pm Post subject: |
 |
|
that usr-merge crap is done by (again) pottering.
look at udev >=236 problem, with udevadm link against a library from /usr https://forums.gentoo.org/viewtopic-t-1076204.html
the usr-merge is an answer to 1/ people making crappy tools (critical tools and datas should not be seek in /usr but only in / ; /lib, /sbin or /etc are perfect for that) 2/ fedora use an initrd, so why not bug everyone with an initrd too?
usr-merge is not a solve to the real problem, the problem is these windows mentality devs that known nothing about linux and code like crap, and instead of fixing their program, they have fucked /usr mount
If all devs use that mentality, then Roy Marples would had install dhcpcd into /boot ; and to fix that mess, he would had make mandatory to mount /boot in an initrd or have /boot always merge in / in openrc
I think this example make it clear what is the problem, and how they "solve" it. |
|
| Back to top |
|
 |
saellaven
l33t
Joined: 23 Jul 2006
Posts: 646
|
| Posted: Tue Nov 13, 2018 2:23 pm Post subject: |
|
|
| krinn wrote: |
If all devs use that mentality, then Roy Marples would had install dhcpcd into /boot ; and to fix that mess, he would had make mandatory to mount /boot in an initrd or have /boot always merge in / in openrc
I think this example make it clear what is the problem, and how they "solve" it. |
and having a separate /usr with openrc was something that just worked before William Hubbs convinced the Council to break it for the benefit of systemd (since Poettering can't seem to conceive use cases outside of his own). Good thing Hubbs is both on the Council (which is why the Council rubber stamped his proposal without doing any investigating of their own and without knowing the Hubbs deliberately didn't inform them about SteveL's patches) and lead of openrc even though he's pro-systemd/anti-openrc.
the initrd/initramfs "solution" is far more prone to breakage unless you rebuild them every time a file included in it gets updated on the base system but not in the initramfs, since not every system package is guaranteed to be backward or forward compatible. |
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Wed Nov 14, 2018 9:35 am Post subject: |
|
|
| NeddySeagoon wrote: | You can have a split /usr without an initrd still.
There are two ways I'm aware of.
1) Throw udev and friends away and go back to a static /dev. No auto anything. That works for me.
2) Start udev after localmount. I haven't done this for a while, so I don''t know if steveLs patches to OpenRC for this still work. |
Work fine here, on openrc-0.23.2 (iirc.)
Although I am using eudev, that should not make any difference (since it's the same codebase, including dumbassery like "stupid network names".) | NeddySeagoon wrote: | | The initrd has become the new root and the merged /usr and root has become the new /usr. |
Weird, I said exactly that several years ago, when this whole crazy idea was mooted; the split works for both sides, in the longer-term, whereas the merge simply makes real-life usage a PITA for zero benefit (apart from assuaging a few of the egos involved.)
OFC none of the "wise-heads" in Gentoo were interested in continued collaboration of all sides; only in pushing ahead with what everyone knew was stupidity ("but RedHatGooglePlex might see me demur.. yikes.") |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Wed Nov 14, 2018 3:54 pm Post subject: |
|
|
| steveL wrote: | | Although I am using eudev, that should not make any difference (since it's the same codebase, including dumbassery like "stupid network names".) |
?????? |
|
| Back to top |
|
|
NeddySeagoon
Administrator
Joined: 05 Jul 2003
Posts: 54220
Location: 56N 3W
|
| Posted: Wed Nov 14, 2018 8:14 pm Post subject: |
|
|
Tony0945,
The context here is
| NeddySeagoon wrote: | 2) Start udev after localmount. I haven't done this for a while, so I don''t know if steveLs patches to OpenRC for this still work.
| steveL wrote: | Work fine here, on openrc-0.23.2 (iirc.)
Although I am using eudev, that should not make any difference (since it's the same codebase, including dumbassery like "stupid network names".) |
|
I tested the "start udev after localmount" patch set some time ago. steveL is confirming that it still works on openrc-0.23.2 and bemoaning the "stupid network names".
_________________
Regards,
NeddySeagoon
Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail. |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Wed Nov 14, 2018 10:38 pm Post subject: |
|
|
Neddy, I should have been more precise. I wanted to know what SteveL meant by stupid network names. It has just come to me though. I'm sure he means names like "empanada" or whatever. I just get rid of those by command line "net.ifnames=0". I'll agree that having the "stupid network names" as default was indeed stupid.
I apologize for the memory lapse. |
|
| Back to top |
|
|
berferd
Tux's lil' helper
Joined: 13 May 2004
Posts: 117
|
| Posted: Thu Nov 15, 2018 3:19 am Post subject: |
|
|
| Anon-E-moose wrote: | | ...The split-usr flag is what I'm talking about. |
Masked. Thanks for the heads up. |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
| Posted: Thu Nov 15, 2018 3:38 am Post subject: |
|
|
| berferd wrote: | | Anon-E-moose wrote: | | ...The split-usr flag is what I'm talking about. |
Masked. Thanks for the heads up. |
Don't forget coreutils.
I added split-usr to USE= in /etc/portage/make.conf, just in case more of these packages show up.
Even if you want a flat file system, I recommend adding split-usr (with or without '-') in make.conf so your flags are consistent across packages. |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Thu Nov 15, 2018 6:15 am Post subject: |
|
|
| Tony0945 wrote: | | berferd wrote: | | Anon-E-moose wrote: | | ...The split-usr flag is what I'm talking about. |
Masked. Thanks for the heads up. |
Don't forget coreutils.
I added split-usr to USE= in /etc/portage/make.conf, just in case more of these packages show up.
Even if you want a flat file system, I recommend adding split-usr (with or without '-') in make.conf so your flags are consistent across packages. |
also prevents a silent change of default fking your system ... Not that Gentoo has done that *cough* ffmpeg
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Thu Nov 15, 2018 3:19 pm Post subject: |
|
|
So, some time back RedHat announced Statis as a "next gen filesystem".
Originally they were backing BTRFS but it was failing to produce. ZFS is not an option due to licensing. (there is bcachefs but not quite yet).
The original idea was to take XFS and extend it (in userland) to realise the functionality deemed needed for such a filesystem. I raised concerns (a few pages back in this thread) that this will probably hook into systemd resulting in a filesystem dependant on such a concept
Red Hat Enterprise Linux 8 Beta has now been released, lo and behold ... the build relies on systemd ( https://github.com/stratis-storage/stratisd/search?q=systemd.h&unscoped_q=systemd.h )
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
steveL
Watchman
Joined: 13 Sep 2006
Posts: 5153
Location: The Peanut Gallery
|
| Posted: Thu Nov 15, 2018 4:18 pm Post subject: |
|
|
| Yamakuzure wrote: | | After re-reading ... well... not my most brilliant moment... |
Heh, we all get them..
With respect, your post seems to be missing my point, which is not so much about the technical, but about the social effects of the Propaganda.
Even though the two points you made in passing at the start of your earlier post have been soundly dismissed on these forums on several occasions, that you yourself were party to, you stated them as if they were givens that no-one could argue with.
Where the split originally came from is irrelevant, again as previously discussed (practical usage is not theory, and if something works for several use-cases in practice, it usually indicates a pivot we should keep.)
As you've pointed out, if the merge is enforced it breaks practical use-cases, and here's the thing: the "traditional" or flexible approach, does not enforce anything on anyone, certainly not in usage. It just requires a bit of discipline from programmers working on early-boot, who usually do this by default, as they normally know the need for flexibility in real-world usage.
It is only recently that we've had web-coders coming in from winbloze and crapple, trying to replicate stupidity in ignorance, as any sort of factor in the Linux demi-monde.
But yeah, my point was not technical so much as sociopolitical, on how insidious the "uniform pattern of public utterances" enforced by Propaganda truly is in reality: it had even you repeating lies as given, even though you knew them to be false. |
|
| Back to top |
|
|
Tony0945
Watchman
Joined: 25 Jul 2006
Posts: 5127
Location: Illinois, USA
|
|
| Back to top |
|
|
Anon-E-moose
Watchman
Joined: 23 May 2008
Posts: 6098
Location: Dallas area
|
| Posted: Thu Nov 15, 2018 11:20 pm Post subject: |
|
|
Well, RH has been wanting to wrap "linux" up for a while now.
Right now, linux is somewhat like dos6/win286/early win3 in that in those days you boot into dos and start windows or have it autostart (like a DM) windows for you, but stil have dos underneath for those that needed or those who didn't need the hand holding.
RH wants to take it to win98/winnt and later versions stage where you can't easily get to the plumbing underneath what you see (the DM).
In the case of NT and later systems, you never see the kernel (dos or micro-kernel) you can only do what they allow you to do by way of their tools.
I think, they think, it'll make linux more popular.
What RH doesn't understand is MS isn't everywhere because of technical merits, it's there because Bill and company understood marketing concepts.
They convinced computer makers to put it on every system sold (for a small fee, in the scheme of things).
Prior to that, you bought the computer, then had to buy an OS to put on it, either DR Dos, CPM86, IBM Dos or MS dos or some other variant, like forth, etc.
IBM was the biggest competitor to MS, in the early OS days, as they had their version of dos and then OS2 to work against Windows, but MS got to all the other manufacturers and the rest is history.
_________________
PRIME x570-pro, 3700x, 6.1 zen kernel
gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
| Back to top |
|
|
tld
Veteran
Joined: 09 Dec 2003
Posts: 1816
|
| Posted: Thu Nov 15, 2018 11:30 pm Post subject: |
|
|
Holy crap...of all the scary shit I've seen around systemd (starting from the first time I read about the binary log fiasco)...that may be the scariest.
The irony on that one is painful too: Follow me on this one, as it's rather amazing: As of CentOS 6.x (and I have to imagine they were lost following the lead of RHEL6)...for reasons I can't possibly imagine...by default they switched to using crony-anacron!...something clearly designed for anything but a server! As I recall that provides no way to actually schedule jobs for specific times! I recall that step #1 after installing CentOS 6 was to replace that with crony-noanacron as the Good Lord intended.
But now if I google "systemd timers" I get countless links about how they make shit so much easier than cron! I mean you can't make this shit up. Actually it's not unlike their other self fulfilling prophecy / propaganda as to how systemd is better than their worst of all possible init scripts. There's my rant for the day....
Tom |
|
| Back to top |
|
|
Morality124
Tux's lil' helper
Joined: 20 Feb 2018
Posts: 102
|
| Posted: Fri Nov 16, 2018 1:18 am Post subject: |
|
|
| Anon-E-moose wrote: | Right now, linux is somewhat like dos6/win286/early win3 in that in those days you boot into dos and start windows or have it autostart (like a DM) windows for you, but stil have dos underneath for those that needed or those who didn't need the hand holding.
RH wants to take it to win98/winnt and later versions stage where you can't easily get to the plumbing underneath what you see (the DM). |
What an excellent metaphor. Systemd (plus Gnome, etc.) is a lot like Windows 3.11, including in terms of heavy-handed/manipulative adoption tactics (such as various subtle dependencies introduced, reminiscent of the undocumented Windows integration functionality silently added to MS-DOS). |
|
| Back to top |
|
|
Morality124
Tux's lil' helper
Joined: 20 Feb 2018
Posts: 102
|
| Posted: Fri Nov 16, 2018 1:26 am Post subject: |
|
|
| Naib wrote: | | Originally they were backing BTRFS but it was failing to produce. |
Not so sure... there was an incident where a systemd developer had shown up in a BTRFS bug report comment section (with the bug being about a race condition) to repeatedly claim that BTRFS was a "device manager" and as such needed it's own "btrfs.service" with systemd in order to fix the bug in question. The BTRFS developers in response told the systemd developer to pound sand and reaffirmed over and over that BTRFS was a filesystem, not a device manager, and that the issue was with systemd's stupid design causing such race conditions in the first place.
I wonder if the BTRFS developers' lack of... erm... confidence in systemd was a large part in Red Hat marginalizing the project. |
|
| Back to top |
|
|
depontius
Advocate
Joined: 05 May 2004
Posts: 3509
|
| Posted: Fri Nov 16, 2018 2:23 am Post subject: |
|
|
| Morality124 wrote: | | ...(BTRFS, whatever)... |
NOTABUG WONTFIX
_________________
.sigs waste space and bandwidth |
|
| Back to top |
|
|
Morality124
Tux's lil' helper
Joined: 20 Feb 2018
Posts: 102
|
| Posted: Fri Nov 16, 2018 3:03 am Post subject: |
|
|
| depontius wrote: | | NOTABUG WONTFIX |
Yep, pretty much:
| Tomasz Pala wrote: | | This is not a systemd issue |
Rule 1): When you introduce something that breaks userspace, it's userspaces' fault.
Correction to my post - it was a mailing list conversion, not a bug report comment section, but the summary is the same nonetheless. Here is the convo: https://mail-archive.com/linux-btrfs@vger.kernel.org/msg72908.html |
|
| Back to top |
|
|
Yamakuzure
Advocate
Joined: 21 Jun 2006
Posts: 2284
Location: Adendorf, Germany
|
| Posted: Fri Nov 16, 2018 7:43 am Post subject: |
|
|
| steveL wrote: | | Yamakuzure wrote: | | After re-reading ... well... not my most brilliant moment... |
Heh, we all get them..
With respect, your post seems to be missing my point, which is not so much about the technical, but about the social effects of the Propaganda.
Even though the two points you made in passing at the start of your earlier post have been soundly dismissed on these forums on several occasions, that you yourself were party to, you stated them as if they were givens that no-one could argue with. |
Exactly. I was merely responding to previous posts, and in particular to the content of the linked article on freedesktop.org. Doing so without any quote or at least any hint was just stupid.
| steveL wrote: | | Where the split originally came from is irrelevant, again as previously discussed (practical usage is not theory, and if something works for several use-cases in practice, it usually indicates a pivot we should keep.) |
Au contraire! It helps understanding their (freedesktop) motivation and the motivation of the Solaris devs before them.
I find their arguments sound and valid, but only for advertising a change of the general defaults. There is no harm in defaulting to a /usr that's located on the root partition nowadays. As several installers, like the Debian one, already do for ages. And with that, a usr-merge means no harm at all.
But that does not mean, that the usr-split is unnecessary per se. But that's what they seem to propagate, and I can't disagree more.
Those installers still allow users to change their partition layout. And I hope that this stays being the case.
...not to mention the ton of other distributions that do their own thing apart what the "big four" seem to dictate...
Well, sometimes it feels like the freedesktop crowd is convinced that "giving users a choice" is some sort of evil that has to be vanquished...
| steveL wrote: | | As you've pointed out, if the merge is enforced it breaks practical use-cases, and here's the thing: the "traditional" or flexible approach, does not enforce anything on anyone, certainly not in usage. It just requires a bit of discipline from programmers working on early-boot, who usually do this by default, as they normally know the need for flexibility in real-world usage. |
Hear, hear!
| steveL wrote: | It is only recently that we've had web-coders coming in from winbloze and crapple, trying to replicate stupidity in ignorance, as any sort of factor in the Linux demi-monde.
But yeah, my point was not technical so much as sociopolitical, on how insidious the "uniform pattern of public utterances" enforced by Propaganda truly is in reality: it had even you repeating lies as given, even though you knew them to be false. |
I hope I have clarified this by now.
_________________
Important German:- "Aha" - German reaction to pretend that you are really interested while giving no f*ck.
- "Tja" - German reaction to the apocalypse, nuclear war, an alien invasion or no bread in the house.
|
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Fri Nov 16, 2018 8:13 am Post subject: |
|
|
| Morality124 wrote: | | Naib wrote: | | Originally they were backing BTRFS but it was failing to produce. |
Not so sure... there was an incident where a systemd developer had shown up in a BTRFS bug report comment section (with the bug being about a race condition) to repeatedly claim that BTRFS was a "device manager" and as such needed it's own "btrfs.service" with systemd in order to fix the bug in question. The BTRFS developers in response told the systemd developer to pound sand and reaffirmed over and over that BTRFS was a filesystem, not a device manager, and that the issue was with systemd's stupid design causing such race conditions in the first place.
I wonder if the BTRFS developers' lack of... erm... confidence in systemd was a large part in Red Hat marginalizing the project. |
that may have been the motivation but in general btrfs has been a disappointment.
I'm just hoping bcachefs doesn't lose its way but doesn't take too long
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
Morality124
Tux's lil' helper
Joined: 20 Feb 2018
Posts: 102
|
| Posted: Fri Nov 16, 2018 8:19 am Post subject: |
|
|
| Naib wrote: | | that may have been the motivation but in general btrfs has been a disappointment. |
Well luckily Red Hat/2 did a "hold your beer" on that disappointment with this new, "memory-safe" filesystem that is totally dependent on systemd. Ze zekurity! |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Fri Nov 16, 2018 10:38 am Post subject: |
|
|
| Morality124 wrote: | | Naib wrote: | | that may have been the motivation but in general btrfs has been a disappointment. |
Well luckily Red Hat/2 did a "hold your beer" on that disappointment with this new, "memory-safe" filesystem that is totally dependent on systemd. Ze zekurity! |
totally  its a race to the bottom and no ones a winner
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
depontius
Advocate
Joined: 05 May 2004
Posts: 3509
|
| Posted: Fri Nov 16, 2018 8:03 pm Post subject: |
|
|
| Morality124 wrote: | | Naib wrote: | | that may have been the motivation but in general btrfs has been a disappointment. |
Well luckily Red Hat/2 did a "hold your beer" on that disappointment with this new, "memory-safe" filesystem that is totally dependent on systemd. Ze zekurity! |
I just had a quick look, and while it's heavily wrapped up in dbus, I didn't see anything about systemd in there.
_________________
.sigs waste space and bandwidth |
|
| Back to top |
|
|
Naib
Watchman
Joined: 21 May 2004
Posts: 6051
Location: Removed by Neddy
|
| Posted: Fri Nov 16, 2018 8:12 pm Post subject: |
|
|
| depontius wrote: | | Morality124 wrote: | | Naib wrote: | | that may have been the motivation but in general btrfs has been a disappointment. |
Well luckily Red Hat/2 did a "hold your beer" on that disappointment with this new, "memory-safe" filesystem that is totally dependent on systemd. Ze zekurity! |
I just had a quick look, and while it's heavily wrapped up in dbus, I didn't see anything about systemd in there. |
the build calls systemd (which could be patched out) and it only provides statisd.service (which is simple so an openrc init file could be made).
_________________
| Quote: | | Removed by Chiitoo |
|
|
| Back to top |
|
|
Anon-E-moose
Watchman
Joined: 23 May 2008
Posts: 6098
Location: Dallas area
|
| Posted: Fri Nov 16, 2018 8:43 pm Post subject: |
|
|
A filesystem that uses dbus *shakes head* what will they think of next
_________________
PRIME x570-pro, 3700x, 6.1 zen kernel
gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
|
Gentoo Chat
