What is your favourite login manager?
- Login o registrati per inviare commenti
For me SDDM of course, that’s very simple, very themeable, Qt-written, X and Wayland compatible. I now have very many of SDDM themes on my laptop using the same shared /usr/share/sddm partition. :)
However for Trisquel, I will instead prefer LightDM due to Ubuntu conventions (I wonder how to theme for LightDM, not that easy I think), but for the remaining like Parabola, PureOS, Devuan, Hyperbola, HBSD all are SDDM. :)
id just use xdm if it didnt tend to screw up dpi (more than any alternative.) and im sure you can workaround that too (or theyd probably bugfix it?) but out of the box (whatever that even means with regards to a dm/lm) i dont consider xdm a great option, its not in the race.
for us init freedom types, lightdm has caused a couple problems (i guess?) but for the most part, people have gotten it working in some of the contexts that matter most. i think we mostly agree on lightdm actually.
dm/lm is almost useless imo. almost-- but still definitely not useless. for people who like a graphical experience (whether theyre ubuntus original demographic or kernel hackers that like a finished desktop appearance) a dm is a component worth maintaining. id like slim if it had one or two more features and better defaults, while lightdm already has more than it needs. and this will sound absurd but seriously, slim needs a better default theme/in debian at least.
note that trisquel is the very CLOSEST thing im willing to consider touching in the debian family. i leaned on debian for years, ive sworn off it-- im experimenting with something trisquel-based at the moment. other debian stuff i will not touch again, though i did recommend one other dw-recognised distro recently-- whose main developer i met in person a couple times.
i guess if i was using kde, id want the lm/dm that goes with it. but any lm/dm from something as heavy as gnome or kde id prefer to avoid, unless consistent theming was a major priority (its usually not. i mean, i still prefer icewm. what do you think "goes with" that?) no comment on sddm, sounds nice, havent had the pleasure. nearly of the impression that its "trisquels own" dm, if im wrong i thought i read that somewhere.
What do you mean by "Ubuntu conventions"? Trisquel could use any display manager. I was using GDM with Trisquel 7 because LightDM did not allow to lock GNOME Shell or to show notification when the screensaver was active. With Trisquel 8, those things work: I kept the default, LightDM.
"LightDM did not allow to lock GNOME Shell"
really a serious question here, whether based on being misinformed or just wrong-- are you really sure its "lightdm did not allow" to lock gnome shell, or is it more like "gnome shell, like many things from the gnome team, is not designed to work with anything other than other gnome components?" because anyone can design that way, but the gnome team seems to have a real penchant for it, or takes a special pride in it.
might get downvoted but really, its not the "design" that bothers me. its the taking (and expressing) pride in things that dont work with anything else even when it would probably be trivial to make them more compatible that makes me "hate" gnome. (the same way i hate companies who produce more pronounced forms of lock-in.) so it is definitely a serious (and philosophical) question and not just a chance to gripe or moan as some (in other communities) might paint it.
i dont expect all software to be great, but i DO expect attitudes towards user freedom to not be antagonistic to the point where it shows up in the design. and this isnt just at gnome either, but i think its relevant. in fact i think its going to be (already is) one of the most important issues free software faces these days. its through these not-just-incidental designs that you can push the user around even within the context of free-licensed software. and yes, it is a pet issue of mine for certain.
for me this isnt one or two isolated examples, more like a trend over years of bad experiences. thats the context my question is asked in: are you sure lightdm is the thing to blame here? (i really dont care much about lightdm itself. maybe, only maybe its the best among a category of applications i think is of mild importance altogether. although my experiences with lightdm over years are mostly positive-- not perfect, but pretty nice.)
are you sure lightdm is the thing to blame here?
I do not want to "blame" any project. Just an hypothesis (I did not search): it may be due to the choice of logind as a dependence, instead of the unmaintained (including at the time of Trisquel 7's release) ConsoleKit: https://www.freedesktop.org/wiki/Software/ConsoleKit/
Trisquel's choice (inherited from Ubuntu, but it did not have to because, upstream, GDM was an optional dependency): https://blogs.gnome.org/ovitters/2013/09/25/gnome-and-logindsystemd-thoughts/
Anyway:
- Trisquel 7 + GNOME Shell + LightDM: cannot lock the screen, cannot have notifications on the screensaver;
- Trisquel 7 + GNOME Shell + GDM: those features work;
- Trisquel 8 + GNOME Shell + LightDM (or GDM, certainly): those features work.
i DO expect attitudes towards user freedom to not be antagonistic
I do not see any freedom issue here: GDM (and GNOME as a whole) is free software. Their choices of system libraries or their compatibility with other programs have nothing to do with freedoms. In particular, you are free to fork GDM.
On a side note, I find it scary that Trisquel 8 ships ConsoleKit by default. No development whatsoever in more than five years cannot be good from a security point of view: https://cgit.freedesktop.org/ConsoleKit/log/
ConsoleKit's developers urge (in bold face and with an exclamation mark) GNU/Linux distributions to switch to logind: https://www.freedesktop.org/wiki/Software/ConsoleKit/
"ConsoleKit's developers urge (in bold face and with an exclamation mark) GNU/Linux distributions to switch to logind"
consolekits developers have taken over countless projects and gradually lumped them together into a single project that only a large monopolistic corporation is realistically capable of maintaining or steering. at least if you find a bug you can submit a fix, so thats better than non-free software.
the solution for software freedom isnt to switch to logind, it is to maintain (if necessary, fork) consolekit. other people are likely to fork logind.
"I find it scary that Trisquel 8 ships ConsoleKit by default. No development whatsoever in more than five years cannot be good from a security point of view"
you cant fix any security flaws in anything, until you find one. it would be a lot scarier (and more meaningful) if trisquel knew of an important flaw in consolekit and shipped it without doing anything.
regarding security, i find it scarier that loginds developers have a "please can we multiply the size of the attack surface of everything by 10" design approach. a not-yet-found security flaw in consolekit, by comparison might as well be a not-yet-found security flaw in any other software.
you can fix security flaws in logind, but the approach taken by its developers dont make those flaws easier to find before they are exploited. also, the developers of logind got a pwnie award ("lamest vendor response") for their attitude about certain security practices shown on multiple occasions-- i dont think thats ever going to happen to trisquel devs unless its over microcode patches. i side with the trisquel devs on that too, though with a bit of hesitation. read alex olivas article about non-free microcode for more. (i dont have a link, unless its in my email.)
"you are free to fork GDM."
you are free to fork consolekit too, but that wont address your concerns about security. and your comment doesnt address my concerns either, it dismisses them with a free license as a solution to all problems. (a popular retort that has some substance, since free licenses solve quite a few problems indeed.)
i am the author of the redix thesis, which states that you can co-opt and disrupt free software development in ways that makes it less free while using free licenses. one way to do that is to take over a lot of vital projects and lump them together into a single project with gratuitous interdependency. its really just an old eee trick, and eee is more difficult when the software remains under free licenses.
all redix really says is that you can accomplish the same things with redesign and project management what eee does with proprietary code. its not a freedom issue, except that it moves free software development farther from autonomy and closer to monopolistic actors.
the larger and more unwieldy the project gets, the more unlikely anyone is to fork it-- even if forked, the likelihood of a viable fork decreases significantly. thus you are encumbered by default with the list of wontfixes and bad (monopolistic) attitudes associated with the new project, and everyone else spends not months but years scrambling to figure out how to work around it. that has already happened, and set back free software development for years. i think thats a freedom issue. either way, something will have to be (and continues to be) done about it.
in short: being free to fork consolekit only addresses your security concerns if someone actually forks it. being free to fork gdm cannot address my concerns that its developers design everything to be completely (and increasingly!) impractical to fork. if you want to call it a "paradox" that free software can have freedom issues, go ahead.
the truth is that free licensing is a really good tool against non-freedom. as gpl 2 shows, a good free license still has an attack surface. legal verbiage is code, and it is next-to-impossible to write code that cannot be exploited. you cannot write a license that guarantees user freedom-- you can only write one that greatly assists it, and gpl2 did that. gpl3 does it better, but is similarly imperfect. note that gpl2 was sufficient without updates for a lot longer than 5 years. it took a long time to work out a viable exploit.
basically: license that allows forking a project < software freedom
gratuitous interdependency (a phrase i didnt coin) is a vulnerability found in software freedom itself, and redix is a broad class of exploits for that vulnerability and some others.
Should I follow your recommendation and use ConsoleKit or follow the recommendation of ConsoleKit's developers to *not* use their software, which they abandoned more than five years ago? For some reason, the second option looks more reasonable!
I will pass on commenting your conspiracy theory.
"Should I follow your recommendation and use ConsoleKit or follow the recommendation of ConsoleKit's developers to *not* use their software, which they abandoned more than five years ago?"
by precisely the same argument, everyone /absolutely should have/ just clicked ok and upgraded from previous versions of windows to windows 10. compatibility, bloat, commercial upgrade protocols be damned, nadella knows best-- just click ok and be happy. thats what reasonable people will do regardless of reasons not to or whatever.
"I will pass on commenting your conspiracy theory."
though commenting on it would be a lot more fair than just calling it a "conspiracy theory." its one step away from name-calling, except youre basically name-calling something i said instead of name-calling me. i certainly never said it was a conspiracy theory (on the contrary, eee is an established business practice that is well documented and even punished by the courts) and you didnt support your claim that its one, you simply asserted it.
but im not gunning for a refutation with greater substance, i am sure youve got better things to do-- i am simply complaining about the quality of your dismissal-- i am not demanding you fix it. the free software community is effectively divided, whether it was a successful eee strategy or a wild coincidence. still, you can name-call that argument if it pleases you. no one stands up for this issue and really expects more than that. im not even offended, its only mild disappointment.
> by precisely the same argument, everyone /absolutely should have/ just clicked ok and upgraded from previous versions of windows to windows 10.
Is it that you consider Windows, a proprietary operating system, to be no worse than logind, a libre program under the GNU GPL?
Of course, you're free to ascribe your own values and priorities. But it seems like you might be assigning your own values, which are at odds with the libre software movement, to the libre software movement. If that is indeed what you are doing, then it seems to me that it's unfair to those of us in the libre software movement who disagree with those values of yours. Am I wrong?
There just aren’t tight or strict deps between login manager and desktops, an example is, SDDM doesn’t switch users if you’re on Xfce4, only LightDM or GDM will work, but still not an excuse to blame SDDM, honestly SDDM is most themable and smallest, you still can theme under just SDDM-KCM without an installed Plasma shell. If you need to be flexible between many themes then you will be an SDDM lover, but should mostly not hate Qt5. Still don’t need to love Plasma before.
GDM is bloated and too much GNOME-dependent, LightDM is small but doesn’t have GUI theming modules, but still not an excuse to blame LightDM and GDM, that no people are forced to be stuck on certain free software packages. As told before you are always free to run any non-tiling wm with LXQt desktop, surely no people are forced to be tied on ancient-looking Openbox.
GNOME lovers will prefer GDM then GNOME haters will prefer SDDM, finally ones don’t like Qt or GNOME will prefer LightDM. That’s just your preference and no people can enforce this against your will.
In short every free software packages are run under your own choice. No free software packages are restricting or forcing sth against your will. :)
"no people are forced to be stuck on certain free software packages."
technically, no people are forced to use windows either. but the whole secure boot thing is pretty rotten. this statement doesnt try to prove anything, its only to tries to get you to slow down and rethink this statement. what does it actually mean that "no people are forced?"
"every free software packages are run under your own choice."
as an isolated choice whether to run or not run, yes-- you can remove any software package. but this appears to prove something else too that it doesnt prove, which is that you arent ever restricted.
theres never been a package you couldnt remove, and theres never been a time when removing a package bore zero probability that you wont lose functionality or the ability to run other things when you remove it. what has changed however, is proportion of functionality lost when you remove things that werent previously vital. in other words, more and more things are gratuitously interdependent. (not a phrase i coined.)
on the extreme (will never happen) end of this interdependency, if you remove any single package it removes (or breaks) all other packages. it is difficult, but already demonstrated in more than one example of free software design, to increase this interdependency where it is not actually needed in a way that creates-- lets call it "free lock-in." call it pseudo-lock-in. call it whatever you want, but it is effectively reducing the choices people have, and this is the problem youre dismissing.
breakages happen, the vast majority are innocuous, but this can be used as part of a strategy (like buying github) to hurt free software on its own soil. eee practices can be adapted from a proprietary software strategy, very gradually into the free software ecosystem. i believe this is happening, and if the software changes dont convince you then there are very specific motives (from the halloween documents to the business strategy of another multi-billion-dollar company) that benefit directly from this. it explains not only why, but what direction (what "choice") this goes towards. none of this is worth categorical dismissal if you care about user freedom.
"No free software packages are restricting or forcing sth against your will. :)"
forcing, perhaps not. but that doesnt mean they arent restricting. this is bordering on wordplay that moves carefully away from the reality of the situation. namely:
1. deliberate moves (eee-like) are being made to push certain people away from some software and towards others. but the real problem isnt this as much as the ways in which this being accomplished.
2. instead of being forced, developers are working to make it increasingly difficult to exercise choice. the evidence is twofold: 1. it is increasingly difficult to exercise choice, even when it is important or necessary to do so (we can quibble about "necessary" later.) 2. if its not deliberate, there are developers and fans who mock users and say they have no choice, it isnt about choice. these developers are in the minority, but so are free projects that try to make software less free.
3. these moves are repeatedly documented and protested, at which point someone decides to waive them away with a goofy semantic argument:
technically there were no forced upgrades to windows 10, they could have just removed their ethernet cable or wifi settings and continued using 7 or xp. i mean we can just quibble all day about the meaning of "forced" and miss the truth that people are being restricted, we can do all that with petty semantic arguments. but thats a silly game because it puts in so much effort to miss the point.
this wont ever change until more people commit to being a little more honest about their debate. its not a game, its a search for truth. getting petty with semantics insults everyone involved-- but some of us (myself) are used to it.
magic banana and i seem to be on opposite sides of this debate, but he still points out some of the new restrictions and incompatibilities that are created between choices. the fact that they are new is relevant, because more options are getting broken and choices are being reduced.
if thats an honest limitation and not an effect due to design choices or project management that exists to create the incompatibility (and push users towards a smaller number of choices) then theres not much to do or say. unless there is evidence of efforts to make this happen in the past, say as part of some antitrust case. thats when people should stop being so dismissive and rethink this. and i didnt mean you were being deliberately dishonest. honesty meaning that youll actually seriously consider it before you just wave it away.
you made efforts to support your arguments which is good, but your post is much heavier on assertion that someone is "just wrong":
There just aren’t
no people are forced
you are always free
surely no people are forced
That’s just your preference
no people can enforce this
No free software packages are restricting or forcing
you said theyre basically just wrong 7 times, and your argument (the part that supports your assertions) is buried in your own assertions. this averages out to just a dismissal of facts he presented, and only to counter a conclusion that wasnt even made (i draw that conclusion, he doesnt.) can we be more real than that? freedom is important, categorical dismissal serves no good whatsoever.
lastly, note that most of the complaints about this over the past few years are aimed at things that are running prior to the user opening the first application: init, dm, desktop, power management and (though it was stopped by torvalds) even the kernel. this isnt just application breakage, if anything at all is being attacked, its being attacked at the roots of our ecosystem. dont be so quick or eager to dismiss. this debate isnt an ongoing attack on free software, its a detailed description of an ongoing attack on free software.
"What is your favourite login manager?"
The one that lets me log in to the computer and get on with my work. :)
I don't know what a "login manager" exactly is. But Trisquel 8's default sounds good for me.
I guess that you run startx to start a desktop. ‘‘Login managers’’ are graphical clients which require sudo and launched after init like SDDM, LightDM, GDM. :)
wait, "which require sudo?" if you remove sudo, what happens?
If you need to manually start the DM, with sudo removed you need to know the root password to start the DM and make further configs. If uncomfortable to the risk brought with the direct root logins you need to reinstall sudo.
I am a translator!
SLiM.
I don't have any, it's a black screen with login and passphrase. Whatever was installed was buggy and probably unnecessary to me.
- Login o registrati per inviare commenti