View previous topic :: View next topic |
Author |
Message |
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Tue Apr 02, 2024 4:23 pm Post subject: Moved from 17.1 to split-usr 23.0 But what should i do next? |
|
|
Hello. The latest news in portage said that everybody should move from 17.1 to 23.0 profile and if moving from openrc profile, move to the split-usr FIRST. Ok i did but should i move to the merged-usr profile next or stay with this one? What the advantages and disadvantages of having merged or split usr? Can someone please explain this or give me a link where i can read further about it? BTW i want to keep systemd away as long as possible so systemd-related things doesnt really matter. |
|
Back to top |
|
|
grknight Retired Dev
Joined: 20 Feb 2015 Posts: 1849
|
Posted: Tue Apr 02, 2024 4:28 pm Post subject: |
|
|
You are not required to do anything. You were using split-usr in 17.1 but it just wasn't called that.
If you want to follow the freedesktop people's advice, then merged-usr is possible but not required (for OpenRC).
There is no benefit or disadvantage to either layout. |
|
Back to top |
|
|
freke Veteran
Joined: 23 Jan 2003 Posts: 1024 Location: Somewhere in Denmark
|
|
Back to top |
|
|
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Tue Apr 02, 2024 4:53 pm Post subject: |
|
|
Thanks. So if i understood it right, merging the usr gonna give more compatibility to NON-Linux programs (like Solaris), but also can cause bugs in Gentoo itself. Well i think its better to stay on the split one, until it will be absolutely required to merge. |
|
Back to top |
|
|
tld Veteran
Joined: 09 Dec 2003 Posts: 1836
|
Posted: Tue Apr 02, 2024 5:19 pm Post subject: |
|
|
Amon2501 wrote: | Thanks. So if i understood it right, merging the usr gonna give more compatibility to NON-Linux programs (like Solaris), but also can cause bugs in Gentoo itself. Well i think its better to stay on the split one, until it will be absolutely required to merge. | If anything it's more the opposite. The split usr actually IS what traditional Unix (and Linux) always had. The merged usr really came from the systemd folks.
Tom |
|
Back to top |
|
|
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Tue Apr 02, 2024 5:28 pm Post subject: |
|
|
But why split-usr didnt became default as it was in 17.1 then? Forcing new users to install with merged? |
|
Back to top |
|
|
tld Veteran
Joined: 09 Dec 2003 Posts: 1836
|
Posted: Tue Apr 02, 2024 6:21 pm Post subject: |
|
|
Amon2501 wrote: | But why split-usr didnt became default as it was in 17.1 then? Forcing new users to install with merged? | I'm not following what you're saying here. The 17.1 profiles ARE split-usr except for those that are expressly named with merge-usr.
EDIT: I misread what you were saying, but I'm still confused as to what you mean by a "default". With the 23.0 profiles, you have to expressly choose the new profile you want. There's no "default" as to split vs merged.
Tom |
|
Back to top |
|
|
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Tue Apr 02, 2024 6:59 pm Post subject: |
|
|
tld wrote: | but I'm still confused as to what you mean by a "default". With the 23.0 profiles, you have to expressly choose the new profile you want. There's no "default" as to split vs merged.
Tom |
Exactly this. Imagine a new user, who dont know anything about merged or split usrs and he is installing Gentoo for the first time. What will he choose when asked about the profile? Probably the less confusing option like default/linux/amd64/23.0 (stable) maybe gnome or plasma. And what he`s gonna get after that? The Merged profile! An opposite than it was in 17.1. Thats what i meant by "default". A trap for new users. |
|
Back to top |
|
|
flysideways Guru
Joined: 29 Jan 2005 Posts: 466
|
Posted: Tue Apr 02, 2024 8:33 pm Post subject: |
|
|
Quote: | Whenever a certain choice is presented the handbook will try to explain the pros and cons of each choice. Although the text then continues with a default choice (identified by "Default: " in the title), the other possibilities will be documented as well (marked by "Alternative: " in the title). Do not think that the default is what Gentoo recommends. It is, however, the choice that Gentoo believes most users will make. |
There is not presently an explanation, in the handbook, of the choice, or its reasons, between split and merged /usr.
The current handbook will lead unaware users to be unnecessarily merged.
The reality is, there is no choice if one desires systemd, but, openrc retains choice.
The question then becomes, is it the intention to be systemd only? That would appear to be the implicit bias of the handbook authors. |
|
Back to top |
|
|
logrusx Advocate
Joined: 22 Feb 2018 Posts: 2233
|
Posted: Tue Apr 02, 2024 8:43 pm Post subject: |
|
|
Amon2501 wrote: |
Exactly this. Imagine a new user, who dont know anything about merged or split usrs and he is installing Gentoo for the first time. |
That means the user has no knowledge of it and may never be to find out because the new user may never need to. You don't need to worry about that too.
flysideways wrote: |
The question then becomes, is it the intention to be systemd only? That would appear to be the implicit bias of the handbook authors. |
There's no question where it doesn't need to be any.
Best Regards,
Georgi |
|
Back to top |
|
|
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Tue Apr 02, 2024 8:56 pm Post subject: |
|
|
logrusx wrote: | Amon2501 wrote: |
Exactly this. Imagine a new user, who dont know anything about merged or split usrs and he is installing Gentoo for the first time. |
That means the user has no knowledge of it and may never be to find out because the new user may never need to. You don't need to worry about that too.
Best Regards,
Georgi |
Actually there is something really bad about this. The logic is simple: if most of the new users will choose the merged profile even without knowing what they are doing, then someday there ll be a "critical mass" of such users. And when it happens, everyone else (the ones who aware and chose split for a reason) will become the minority. And their needs MAY be abandoned by the maintainers, just because of that. So Gentoo could become just another one systemd-only distro in a future. Thats why a choice between split or merged and between systemd and openrc profiles should be explained clearly in the handbook. |
|
Back to top |
|
|
dmpogo Advocate
Joined: 02 Sep 2004 Posts: 3391 Location: Canada
|
Posted: Wed Apr 03, 2024 12:46 am Post subject: |
|
|
I am with Amon2501 on that. There is obviously a change in default. Default is what happens if things are not explicitly spelled out or set. 23.0 goes merged profile route in this. Which does have repercussions for the future. So it would be nice to read about rationale. |
|
Back to top |
|
|
logrusx Advocate
Joined: 22 Feb 2018 Posts: 2233
|
Posted: Wed Apr 03, 2024 3:55 am Post subject: |
|
|
Amon2501 wrote: | And when it happens, everyone else (the ones who aware and chose split for a reason) will become the minority. |
They are minority now.
Now if you'll excuse me, I find this conversation meaningless.
Best Regards,
Georgi |
|
Back to top |
|
|
pjp Administrator
Joined: 16 Apr 2002 Posts: 20421
|
Posted: Wed Apr 03, 2024 4:10 am Post subject: |
|
|
systemd adoption has already reached critical mass. Not using systemd is already the minority.
I don't speak for developers, but it seems that the only reason Gentoo isn't a systemd distro is because there are developers who want to use other init systems.
I would frame any request (bug report or wiki discussion) at improving the documentation around the choices available to a user rather than "Something might happen that I don't like." Were I a developer, I'd rather see the former than the latter. If they're driven away, then surely the available choices will go with them.
(Disclaimer: I use a 17.1 profile, openrc, and "split usr".) _________________ Quis separabit? Quo animo? |
|
Back to top |
|
|
flexibeast Guru
Joined: 04 Apr 2022 Posts: 399 Location: Naarm/Melbourne, Australia
|
Posted: Wed Apr 03, 2024 5:12 am Post subject: |
|
|
i'm not a systemd user, and will move distros rather than use systemd (as i did when i moved from Debian to Void several years ago), for various reasons. Nevertheless, i recently moved to a merged-usr profile, because:
(a) it's primarily a historical artifact (which i say as a Unix history enthusiast);
(b) i have no specific concrete use-case that prevents me from doing so; and
(c) decades of experience with Linux means i try to minimise the extent to which my system "lives outside the mainstream".
This is the primary reason i've been actively moving to Wayland - not because X wasn't doing what i needed (in recent times it mostly was), but because it seems to me that the writing is on the wall for X: Matthieu Herrb, an Xorg dev and a primary OpenBSD dev in this area, did a presentation last year [PDF] saying "X11 is fading away" and "Wayland is the way to go for graphical desktop", in the context of a platform known for being careful in what it takes on (i maintain a couple of OpenBSD servers myself).
So although i'm willing to actively do work to support the continued existence of systemd alternatives - cf. my work porting the s6 ecosystem documentation to man pages - there are various things that are not actually part of systemd (even if used by it) that i don't particularly see any point fighting against. There are lots of people who express unhappiness with how the systemd world is impacting the Linux ecosystem, but far fewer for whom it's important enough that they roll up their sleeves and take on some work themselves.
(D-Bus is another example. Even though i'm not a fan of its design, it meets a genuine set of needs - cf. the introduction to my "D-Bus: The Essentials" guide - and more specifically, i do make use of functionality it provides. For example, as things currently stand, Bluetooth headphone support requires the use of D-Bus.) |
|
Back to top |
|
|
Amon2501 Tux's lil' helper
Joined: 12 Dec 2018 Posts: 140
|
Posted: Wed Apr 03, 2024 11:11 am Post subject: |
|
|
flexibeast wrote: | i'm not a systemd user, and will move distros rather than use systemd (as i did when i moved from Debian to Void several years ago), for various reasons. Nevertheless, i recently moved to a merged-usr profile, because:
(a) it's primarily a historical artifact (which i say as a Unix history enthusiast);
Quote: | (b) i have no specific concrete use-case that prevents me from doing so |
|
These are. But still are there any use cases for moving to merged profile besides systemd stuff and just "being mainstream"? Something like "in merged profile <something> will run better than in the split one"? |
|
Back to top |
|
|
flexibeast Guru
Joined: 04 Apr 2022 Posts: 399 Location: Naarm/Melbourne, Australia
|
Posted: Wed Apr 03, 2024 12:16 pm Post subject: |
|
|
Amon2501 wrote: | are there any use cases for moving to merged profile besides systemd stuff and just "being mainstream"? |
Well, by "being mainstream" i intended to imply that, based on my experiences, i'll be less likely to increasingly run into problems, of varying sizes, due to having a configuration that's different from the one assumed by an increasing number of packages/projects, and increasingly 'unsupported'. That's a big advantage for me: i've got way too much else on my plate to be trying to maintain a certain configuration not because i actually need it, but simply out of some sense of sheer bloody-mindedness[a]. And the 'cost' of moving to merged-usr, which i'd already done prior to the release of the 23.0 profile, was running the relevant script, which completed in less than a few minutes (i can't remember the exact time). So far i've not noticed any negative impacts, but of course others' mileage may vary.
i certainly don't consider "being mainstream" to be a virtue in itself!
That said, i'd be interested in any answers others might have in response to your question.
[a] Or reflexive opposition to anything that even looks like it might be systemd-related. My favourite examples of the absurdity this can reach are:
* someone assuming sysctl(8) was a systemd thing. No; it was introduced by 4.4BSD in the mid-90s.
* someone assuming anything with the `-d` suffix is a systemd thing. No; it's been used to mean `daemon` for a long time. i ended up asking The Unix Heritage Society list about it; the first use in Unix was in 1972, with 2nd Edition's `dpd` ("data-phone daemon"). But the suffix was also used in systems that predated Unix. |
|
Back to top |
|
|
dmpogo Advocate
Joined: 02 Sep 2004 Posts: 3391 Location: Canada
|
Posted: Wed Apr 03, 2024 8:02 pm Post subject: |
|
|
flexibeast wrote: | Amon2501 wrote: | are there any use cases for moving to merged profile besides systemd stuff and just "being mainstream"? |
Well, by "being mainstream" i intended to imply that, based on my experiences, i'll be less likely to increasingly run into problems, of varying sizes, due to having a configuration that's different from the one assumed by an increasing number of packages/projects, and increasingly 'unsupported'. That's a big advantage for me: i've got way too much else on my plate to be trying to maintain a certain configuration not because i actually need it, but simply out of some sense of sheer bloody-mindedness[a]. And the 'cost' of moving to merged-usr, which i'd already done prior to the release of the 23.0 profile, was running the relevant script, which completed in less than a few minutes (i can't remember the exact time). So far i've not noticed any negative impacts, but of course others' mileage may vary.
i certainly don't consider "being mainstream" to be a virtue in itself!
|
Well, there is certainly less issues with X that with wayland for now, so you moving actively to wayland shows that you do not seek a quiet life all the time
Saying that, I wonder what is a difference between 17.1 and 23.0 profiles. Is there a point moving ? |
|
Back to top |
|
|
pietinger Moderator
Joined: 17 Oct 2006 Posts: 4892 Location: Bavaria
|
Posted: Wed Apr 03, 2024 9:29 pm Post subject: |
|
|
dmpogo wrote: | Saying that, I wonder what is a difference between 17.1 and 23.0 profiles. Is there a point moving ? |
I dont know all differences, but for me the most important is Security ... e.g. look at the use-flags NOW default on for gcc ... _________________ https://wiki.gentoo.org/wiki/User:Pietinger |
|
Back to top |
|
|
flexibeast Guru
Joined: 04 Apr 2022 Posts: 399 Location: Naarm/Melbourne, Australia
|
Posted: Wed Apr 03, 2024 11:56 pm Post subject: |
|
|
dmpogo wrote: | Well, there is certainly less issues with X that with wayland for now, so you moving actively to wayland shows that you do not seek a quiet life all the time :) |
[ i'm writing looong comments about this stuff because it's increasingly been on my mind, and this thread has merely provided the opportunity to get it off my chest. ]
Well, i do want a quiet life, but "everything flows", the world keeps moving on, and for me the question is: How do i choose to respond to that?
i've seen many changes in the Linux ecosystem over the years. i feel i've developed some sense of how certain areas are changing, and to what extent it might be possible to change a given trend and/or its trajectory.
There's been so much anger and vitriol expressed about systemd. Has that significantly slowed the systemd juggernaut? Not really. Not least because, as in the case of D-Bus, and as in the case of Wayland, it addresses very real issues for significant numbers of people.
For example: unlike on, say, OpenBSD, which has developed a pretty clean shell-script-based service management system, with a 'standard library' in the form of rc.subr(8), the situation on Linux was a mess. Many of the (usually volunteers) who maintain packages for Linux don't want to have to learn the complexities of shell scripting and the subtle issues that can arise, or develop and maintain workarounds for race conditions, and so on. systemd comes along and says: "Hey, with systemd, you'll be able to write service definitions declaratively; you won't need to wrangle shell scripts." That's a pretty attractive proposition to a number of package maintainers, and in the absence of systemd alternatives explicitly providing such an interface[a] - not just saying "oh that could be done on our alternative" - those maintainers are going to be inclined towards systemd, regardless of what design and implementation issues are involved in systemd's approach.
So in wanting to try to ensure that myself and others have choices and alternatives available, i feel that ranting against the incoming tide, like a tech King Cnut, is typically far less effective than actually putting in the work to develop and support those choices and alternatives. Which, for me, includes documentation work: i wrote my "D-Bus: The essentials" guide because i wanted to provide people with information to help them make an informed choice about whether they want to use it, and how - at the time i put it together, D-Bus info seemed to be mostly either "You need this for desktop stuff; shut up, sit down, and enjoy the in-flight movie" or "Here's everything you never wanted to know about D-Bus, and more".
Does this create more work for me in the short term? Definitely. But my hope is that it's a longer-term investment, in two ways:
* i'm making the changes at my own pace, on my own terms, rather than possibly being forced to do it at a later point at a time that's significantly more inconvenient.
* Helping others to configure and maintain non-'standard'/non-'mainstream' setups might help create a larger community for such setups, and maybe at least some of those people might do work that will, in turn, support me configuring and maintaining my own non-'standard'/non-'mainstream' setup.
To address your comment more specifically: moving to a Wayland-based system hasn't resulted in any problems that have been due to Wayland itself. (i actually started looking into Wayland again in recent times because i wanted to check where things were for myself, instead of just taking all the bluster about it at face value.) However:
* i don't do graphics-intensive stuff. i'm basically not a gamer, and streaming video is the most graphics-intensive thing i do. Related to that:
* i don't have an Nvidia-based system, by choice, having for many years observed Nvidia's, mm, 'unsupportive' attitude towards the FOSS world. Nvidia is known for making things unnecessarily difficult for FOSS devs trying to get stuff working on their hardware.
Still, it's certainly true that moving to Wayland has involved work i'd not otherwise be doing, such as developing new configs as i move to new software, and wrangling issues that have arisen in doing so. But i'm documenting what i'm learning as i go - e.g. in setting up Waybar, to replace my X11 Polybar setup, i created the the 'Waybar' page on the wiki.
[a] The '66' system, used by the Obarun distro (an Arch variant), provides an example of a declarative approach to service definition can be layered on top of s6. i used s6+66 when i was using Void as my daily driver.
dmpogo wrote: | Saying that, I wonder what is a difference between 17.1 and 23.0 profiles. Is there a point moving ? |
The relevant News item says:
Quote: | The new 23.0 profiles enable some toolchain hardening features and performance enhancements by default, and standardize settings. You can find the list of changes on the wiki tracking page [1]. |
|
|
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
|
|