Comparison of bootloaders

It is my personal logical conclusion of the software used in most distros. Eg. Most have adapted to SystemD as system manager.
Most distribution maintainers like to minimize the required software for an operational distro.
Because systemd-boot (sd-boot for short) is part of SystemD, it will become the default bootloader on systems that use SystemD in the long run.
If you guys insist on source references you can use my replies on the internet mentioning it :rofl:
So no it is not an “official statement” by Linus Torvalts (who else would you accept as official source right?), it is just a logical deduction of what and how software is used in the community…

Nowadays that’s correct (your statement) but in the early stages of Linux it was the only one acceptable for multiple hardwares.

Thank you, but no i never force my ideas i only try to persuade by using logical arguments, which is all anyone can and should do on the internet. (IMHO)
Yes i might come over a bit hard headed, but that’s a flaw of my logical mind that has problems with non-logical solutions. (Or ways of doing things)
If you have noticed most of the times i make statements i use “IMHO” which we all know what it means right? :smiling_face_with_three_hearts:
(Sometimes i forget to use it, but its always in my intentions when i write stuff, im human afterall)

Not at all, but its human that people perceive it that way when their opinions are being questioned.
I never intend to hear just the words “You’re right”, even in cases that i might be i prefer arguments explaining why im right or wrong.
(That’s the only way every human will learn new things and correct wrong info it had)

It’s a ventilation of frustration from an old hacker that coded in assembly in the past and noticed how new “Programmers” created new code.
Plus 2 questions to make him think (and share with us) about where Manjaro is heading…

And no it stays M$ for me and most others who know the reason of it…
(I refuse to write that name in full)

All in all: Yes im an old fart, take me with a bit of salt, im really pleasant when you get to know me, and my intentions are always “good”

PS: Tijd voor een bakkie koffie :smiley:

3 Likes

I believe it would be better to have an option in the installer to select the boot loader and init.
let the users decide what they want. It would also simplify the process.

1 Like

No such luck I’m afraid.

I’ve been trying to ask you what the benefit you foresee is. For some reason you seem to be disinclined to answer this, or maybe you just feel that the extra performance is worth it? I like good performance, but there is a limit on how how much time I would use to save 0-10 seconds a day, when that time comes out of my sleeping hours. My backlog is long enough as it is. Again, if someone else codes the ecosystem around it, it becomes much more interesting proposition. But I don’t have resources to do it myself in the foreseeable future.

Legacy boot support is not for me. I’ve actually never used anything but uefi boot on my own linux systems. But as a distro, we need to consider also other users and not just think about ourselves.

Just because systemd keeps bundling in functionalities doesn’t mean that it should be used for everything. Should we also switch from networkmanager to systemd-networkd just because it comes with systemd?

2 Likes

So it is an opinion stated as a claim… Mhhokay…
And a source not it does not have to come from Linus that’s bullocks.

I use Linux al long time already so I know about those ‘early stages’ but also in those ‘early stages’ there where alternatives and I am not only talking about (e)Lilo.

Which seems to be a logical solution in you mind does not tent to be logical in all cases. Some logical solutions also may sound valid but can be hard to implement or explain to users.

I know the reason behind that, but still… ahh well it is indeed time for some coffee.

PS. Niets persoonlijks hoor, maar je denkwijze kan ik volgen. Je communicatie daarin echter niet.
(nothing personal, I can follow your way of thinking. But not way you are communicating)

There is. Manjaro-architect supports systemd-boot, refind and grub. Calamares has option to not install bootloader, and systemd-boot and refind installation is usually just one simple command.

3 Likes

I know that in architect we can select boot loader of our choice, i was suggesting to have the same option in other variants. A little more flexibility rather than complete DIY.

We already have an option to select kernels in non architect variants. If you have faith in your users to know better and select kernel for themselves then why not boot loader and init.

There are merits of both systemd-boot and grub so why not have option during installation and let GRUB be the option selected by default for the users who don’t know the difference.

The benefit is easy: Innovation, size and simplicity
THE philosophy of Unix program is “Do one thing and do it as perfect as you can for the problem you try to solve”
Every bit of decrease in execution time or resource usage (RAM or disk wise) is a pro for any program.
And yes i know the burden of a coder, we are slaves of the community whether we like it or not that’s our nature…
I can understand that you don’t have enough time or resources to do it yourself, but instead of trashing ideas because of that, you could give incentives so others who like it might put in the efforts to make it happen.

Exactly my point, the hardware industry is moving so fast that “Legacy Boot”, as the name it self already says, is put in the past already.
So why hang-on to things that won’t be used in the future for software that is meant to be used in the future?
(It’s like trying to keep your C=64 running because you don’t want to get rid of it out of nostalgia…)

SystemD is bundeling a lot, yes i know, and no it doesn’t mean you have to use them all, but it sure makes things easier from a pov of maintanability and distro-agnostic software.

When it comes to NetworkManager vs systemd-networked, i already do that and find it much better.
The only reason i keep NetworkManager around is because of the nice GUI icon to access WiFi-AP’s around my machine…(Which i never use anyhow :smiley: )

That’s were the “debateing” comes into view…
One side of a topic tries to persuade the other using arguments, the strongest arguments win at end.
(Strongest not meaning the one shouted most, but instead the one acceptable to most)

PS: (Lets keep it English for the rest, i only made a personal wink to let you know where i grew up :wink: )
The reason why my communication could be hard to follow is because English is not my native language in mind.
It might look like it, but that’s the years of usage of it on the internet…
(I have many dictionaries in my mind)

Keep in mind that with the current sd-boot option it will wipe previous entries, making other distros installed on your system unable to be used…
That needs to be fixed, but the package maker refuses it as i stated in this reply
So although i fully agree with you, i would prefer it to stay as is until the above is fixed, else new users will burn down buildings…

Welcome to the saltiest conversation of the internet, ladies and gentlemen!

Who will win the elections? Gummiboot (systemd-boot) or Grub?

The debate is intense, the results may be stunning, who knows!
Speechless am I, this is a major step in the human history. Neil Armstrong thought he made a major step for the humanity, but that bootloader change will dethrone that magnificence.

I see someone in the audience raising his hand, ready to ask a game changing question : what if you provide checkboxes featuring bootloader options during installation?

Hmmm, this seems like an interesting question. Will we ever know?

coming soon

7 Likes

Unix philosophy is nice, but you also need the other other tools in the ecosystem to work. Currently there aren’t mature tools built around systemd-boot.

Which is the extent of my point. Judging by the adoption rate of sd-boot, this is the case for many other maintainers as well. But as said, I’m open to ideas if there is an upstream doing the development. I don’t mean to trash the idea in general.

Please elaborate.

I also like networkd and use it myself. But it’s entirely different matter to distribute it. I spent some weeks on trying to implement it in bspwm edition as poc, but had to conclude that I could not make it beginner friendly enough to deploy as default on Manjaro. When I setup a simple installation for myself, I might well use systemd-boot and networkd for myself. But that doesn’t align with the goals of Manjaro as a distro. We aim for out of the box experience and supporting a large user base with varied hardware and experience levels.

That’s all part of that same philosophy by nature, because every bit uses the others it needs to perform it’s own task at the end…

Many if not most “upstream” software starts off as a distro-specific solution…
So instead of pointing towards “higher-up” you could give incentives to embrace solutions made for our distro and help it become distro-agnostic as much as can so it could be used as “upstream” for the other distro’s :wink:
(See for example the solution of signed SHIM as used in ubuntu wrt SecureBoot)

You lost me there as i don’t know those “bspwm” and “poc” terms.
Not sure what and how you tried to do what you tried, but im sure with enough exposure it would draw attention from others that might be able to aid you.
(Gotta love GitLab)

Oh please no! And no for “homed” as well :grin:

2 Likes

Any reason for that opinion?

What kind of incentives are you talking about in practice?

Many if not most solutions in community maintained distros like Manjaro start with someone coding something for themselves and then sharing with others. Someone needs to do it first.

I tried to write a simple tui/dmenu interface to connect to a wifi using systemd-networkd. I concluded that it was too much work for the expected benefit.

Carefully read the complete quote i replied to, i quoted those parts to point what you’re doing wrong (imho) and explained how you could do better in that respect.

That is how every piece of software is written, no matter who maintains the distro.
(A distro is only a collection of community crafted software glued together)

Exceptions are M$, :apple: and the like who are not coded by the community at all ofcourse.

Then i would suggest you to dig into how NetworkManager does what it does… (It’s tremendously big)
Like i said in some other thread, the effort needed to make something desired happen should never outweigh the end result…
The efforts put in is only affecting the ones making those efforts, but the result is used (and appriciated intensely) by the community (them self included)

I did. I reread it now. It still don’t understand what you mean :sweat_smile:

I understand what you don’t want me to do (trash systemd-boot or other projects). Instead I should provide “incentive”. I’m asking: what kind of incentive you have in your mind here? I’ve been trying to ask this twice now, but it seems we are not understanding each other unfortunately. :neutral_face:

Considering the language you use of other projects, it is surprising to me that you feel that I’m trashing ideas. My point is that it needs more work. And it doesn’t seem offer much in trade for that work. So someone needs to put in the work for it to be viable alternative.

1 Like

What i mean is that you should not demotivate (new) ideas, but instead reply in such a way that it motivates the writer, or the readers, of it to improve it.
That includes not using arguments like “Too much work for the benefit” or things among the lines of “not our concern and mentioning ‘upstream’ as a solution point”.

To make it more clear and explicit: I’m not against “trashing” something as long as it is accompanied with arguments explaining the reason…

At least we agree on that, but your intention was not that clear to me yet.
I’m glad we’re at same frequency now :+1: :smile_cat:

I enjoy systemd-boot! :smiley:
It is a sweet little boot loader :hugs:

Smile to shine on your beautiful day! :blush:

3 Likes

If you are after speed and has corebootable HW, Kexec payload will grant you the bootspeed of Gods.


For EFI I use rEFInd because it’s really small and finds efi entries automatically
For BIOS systems I use syslinux, I find it easy on Arch linux where the kernel numbers doesn’t change. Not sure what hooks you would need to craft for Manjaro.

I would love that, but unfortunatly im using an ASUS X99 Motherboard.