View previous topic :: View next topic |
Author |
Message |
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Mon Nov 24, 2014 8:34 pm Post subject: |
|
|
Ant: without going back into the code, what's wrong with simply rewriting it to catch the signal, as you want?
WRT setting NULL envvar, that means: clear it. |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Tue Nov 25, 2014 1:51 am Post subject: |
|
|
steveL wrote: | Ant: without going back into the code, what's wrong with simply rewriting it to catch the signal, as you want? |
I think you're misunderstanding - I didn't want sysvinit to catch signals; I wanted sysvinit's tools (which I can't easily avoid because everything is hardcoded to use them) to work under another PID1 that doesn't speak /dev/initctl. In the process of getting that to work I learned about this weird message-passing ping pong ritual, and looked a bit more through the code and realised: most of sysvinit doesn't need to be in init at all, not just the thing I was working on.
What I really wanted to say (before I pulled the topic way off on this tangent) is: sysvinit *is* ugly and stale, but people seem too quick to dismiss it just on account of that, even though it's a fixable thing, and then write entirely new and/or incompatible replacements.
...but I'm still rambling about the complete opposite of the thread subject, so I'll shut up now |
|
Back to top |
|
|
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Tue Nov 25, 2014 12:24 pm Post subject: |
|
|
Ant P. wrote: | I think you're misunderstanding - I didn't want sysvinit to catch signals; I wanted sysvinit's tools (which I can't easily avoid because everything is hardcoded to use them) to work under another PID1 that doesn't speak /dev/initctl. In the process of getting that to work I learned about this weird message-passing ping pong ritual, and looked a bit more through the code and realised: most of sysvinit doesn't need to be in init at all, not just the thing I was working on. |
Well my point was that if we rewrote it to catch the same signals, then shutdown.c could work for both sysvinit or runit (or w/e) as pid1.
However I'd prefer STOP for shutdown, and CONT for reboot (since it makes more sense); then we could collaborate with other non-borg initsystem authors to get that as a standard, so we can share shutdown and reboot code/actions.
Quote: | What I really wanted to say (before I pulled the topic way off on this tangent) is: sysvinit *is* ugly and stale, but people seem too quick to dismiss it just on account of that, even though it's a fixable thing, and then write entirely new and/or incompatible replacements. |
Agreed. Let's also agree that it's established, and get on with improving it. Not by removing functionality, rather by removing code.
Quote: | ...but I'm still rambling about the complete opposite of the thread subject |
Well there's been several threads about the "subject" and one thing that's consistent is the above established ground, and that there's plenty more fun to be had simply by cleaning up what we've got, instead of throwing it away.
So let the convi continue where it continues; it's not like the "politics" haven't been discussed to death, and still are; they're just a bit boring once you've understood the con.
Mods can always split a thread out, if they feel like it (ie: SEP ;) |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Tue Nov 25, 2014 5:31 pm Post subject: |
|
|
steveL wrote: | However I'd prefer STOP for shutdown, and CONT for reboot (since it makes more sense); then we could collaborate with other non-borg initsystem authors to get that as a standard, so we can share shutdown and reboot code/actions. |
That sounds much more sensible than runit's method of CONT plus a file stat. You can't reboot that if the filesystem gets wedged read-only...
sysvinit seems to currently handle those signals (in the same way as bash), but I'm not so sure being able to pause/unpause the init process is a useful thing anyway. |
|
Back to top |
|
|
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Wed Nov 26, 2014 10:34 am Post subject: |
|
|
Ant P. wrote: | sysvinit seems to currently handle those signals (in the same way as bash), but I'm not so sure being able to pause/unpause the init process is a useful thing anyway. |
The only possible use I could see for it is during development, so it behaves as a normal process; this is similar to the "double-fork to avoid a controlling tty" metaphor which is again The Wrong Thing in normal usage (since it means supervision requires pid files instead of simple waitpid.) IOW I agree we should ignore it as a valid use-case for pid 1 (including in a container.)
There is no use-case for shell job control in this context that we should worry about, whereas STOP and CONT for reboot is clearly viable or runit wouldn't be using the latter in any sense. And it'd make things a lot cleaner across the board.
What other uses is /dev/initctl put to? I'd never even heard of it before this discussion. |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Thu Nov 27, 2014 2:57 am Post subject: |
|
|
There's the variables-setting facility, runlevel changing, commands to halt/reboot, and also the thing where shutdown/reboot have delays, the delay amount is sent over that pipe. As is any command to cancel a delayed shutdown.
I may have missed stuff, but that's already a kind of crazy mixed list of things. |
|
Back to top |
|
|
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Sat Nov 29, 2014 4:35 pm Post subject: |
|
|
Ant P. wrote: | There's the variables-setting facility, runlevel changing, commands to halt/reboot, and also the thing where shutdown/reboot have delays, the delay amount is sent over that pipe. As is any command to cancel a delayed shutdown.
I may have missed stuff, but that's already a kind of crazy mixed list of things. |
Hmm I'm tempted to say "if it ain't broke." ;)
I can't see any real use for the variables, given that we can send a delay some other method. Although there is the possibility of setting a variable for new getty sessions, I'm not sure anyone even does that. It just has to be borne in mind as something currently possible, per the spec.
POSIX has had sigqueue() since the early 90s, though you have to use an RT signal to guarantee the queuing and afair the info facility (in the handler.) The latter would be of use to us for setting runlevel, and delay amount, though it might be cleaner to have two separate RT signals for delayed shutdown and reboot, with a value less than or equal to zero being immediate (unless you have a use for negative values). And one to cancel, makes 4.
POSIX guarantees 8, and glibc uses 2 afair, so we should be fine. (Linux has more than 8 ofc, usually 32 though not always that many.)
We also have all the job-control signals available: STOP CONT TSTP TTIN TTOU (might want to use the latter for suspend/hibernate actions.) HUP is reload and not really appropriate to us, though it might be something we want to pass down.(?) kill -l
I'd like us to use USR1 and USR2 for parent/child signalling (USR1 going to child since that's most common situation.) It would be easy enough to then cascade the signals down the tree, for whatever usage we're putting them to in our app-space (most common is synchronisation, ie TELL_PARENT/WAIT_CHILD and vice-versa. So the signal, which the other process is waiting for at that point in the "protocol", means "go".)
Still, if we have to support the variable setting, it'd be difficult to get rid of /dev/initctl (except by reinventing the same mechanism.) What do you think? |
|
Back to top |
|
|
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Mon Dec 01, 2014 6:18 pm Post subject: |
|
|
WRT the nominal subject of this thread ;) this is an interesting email from a while back. Rather prophetic. |
|
Back to top |
|
|
Shamus397 Apprentice
Joined: 03 Apr 2005 Posts: 218 Location: Ur-th
|
Posted: Fri Dec 05, 2014 9:45 pm Post subject: |
|
|
There is a mountain of gold in that thread; just digging back a little bit you find this gem:
Lennart Poettering wrote: | It was clearly a bad idea to post about rtkit on lad. It is a big waste of time fighting this through against all those desktop-haters, fdo-haters, dbus-haters, who apparently believe I am out to take away their freedom to run their systems the way want. I really don't need this shit. The amount of constructive criticism in this thread is next to zero, nobody even bothers to read the README before just fudding around. |
The astute observer will notice that the rhetorical devices are already in use here: the 'foo-haters', the 'bar-haters', the 'baz-haters'. Anyone who doesn't buy into Mr. Poettering's "vision" is therefore a hater of some kind; Q.E.D. Seems the more things change, the more they stay the same... |
|
Back to top |
|
|
khayyam Watchman
Joined: 07 Jun 2012 Posts: 6227 Location: Room 101
|
Posted: Fri Dec 05, 2014 11:42 pm Post subject: |
|
|
Shamus397 wrote: | The astute observer will notice that the rhetorical devices are already in use here: the 'foo-haters', the 'bar-haters', the 'baz-haters'. Anyone who doesn't buy into Mr. Poettering's "vision" is therefore a hater of some kind; Q.E.D. |
Shamus ... its a strategy direct from the propagandists playbook ... take this quote for example:
Joseph Goebbels wrote: | In the newspapers there is insulting and stirring up hatred. Those irresponsible daubers! [In den Zeitungen wird gehetzt und geschimpft. Diese verantwortungslosen Schmieranten!] |
Reading this you might think that every newspaper at the time was directing "criticism" toward the NSDAP, but there was really only one, the Münchner Neueste Nachrichten, most others were either in thrall to the parties meteoric rise, or nullified (via fear and/or timidity). Its cleaver in that its payload is set to tar any disagreement as "insulting", "hateful", etc. Additionally there is a thinly veiled warning there not to write anything that the party might deem "insulting" or to be seen as "stirring up hatred". Given the distance afforded us its not hard to guess who was doing the most to "stir up hatred", in fact you probably only need turn the page of Der Angriff (The Attack) to see exactly that ... but who might make that connection, and voice it, without penalty of "stirring up hatred".
best ... khay
note: this post is on the *nature* of propaganda ... no comparison is made between LP and JG. |
|
Back to top |
|
|
WWWW Tux's lil' helper
Joined: 30 Nov 2014 Posts: 143
|
Posted: Sat Dec 06, 2014 3:55 am Post subject: |
|
|
"THE POLITICS OF SYSTEMD"
systemd isis terrorists.
^^^ nuff sed. |
|
Back to top |
|
|
khayyam Watchman
Joined: 07 Jun 2012 Posts: 6227 Location: Room 101
|
Posted: Sat Dec 06, 2014 9:04 am Post subject: |
|
|
WWWW wrote: | systemd isis terrorists. |
WWWW ... if you can't say something intelligent ... don't say anything at all. Besides the analogy being totally inaccurate (in fact it functions in the same way I outlined above) your simply providing the means for proponents to say "we are being compared to terrorists".
best ... khay |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6145 Location: Dallas area
|
Posted: Sat Dec 06, 2014 6:43 pm Post subject: |
|
|
Was doing some reading (about init systems (thanks shamus), leading to an ignorant guru post about gnome 3) and saw this.
I think it pretty well sums up what is wrong with a few things, sysd, gnome's path, RH's push, etc.
Quote: | In the rush for Linux to become ‘popular’ and ‘make it into the desktop market’, maybe there is an unintended consequence. Not only are Windows users moving to Linux, but Windows devs seem to be arriving as well, bringing their diseases with them – corporate ‘kill off the competition’ mentalities that don’t serve Linux, merely exploit it. |
_________________ PRIME x570-pro, 3700x, 6.1 zen kernel
gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
Back to top |
|
|
UFeindschiff n00b
Joined: 18 Jan 2014 Posts: 25
|
Posted: Sat Dec 06, 2014 11:47 pm Post subject: |
|
|
khayyam wrote: | WWWW wrote: | systemd isis terrorists. |
WWWW ... if you can't say something intelligent ... don't say anything at all. Besides the analogy being totally inaccurate (in fact it functions in the same way I outlined above) your simply providing the means for proponents to say "we are being compared to terrorists".
best ... khay |
well... maybe it's just a throwaway account from Lennart to have a "proof" that systemd is being compared to terrorists |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Sun Dec 07, 2014 3:59 pm Post subject: |
|
|
khayyam wrote: | WWWW wrote: | systemd isis terrorists. |
WWWW ... if you can't say something intelligent ... don't say anything at all. |
There's no doubt he came here intending to be an obnoxious asshole (the other 28 "W"s were removed a few minutes after he showed up - thanks to whoever fixed that). I'm wondering how he's not banned already with that attitude problem, since there's no evidence that it's going to improve any time soon. |
|
Back to top |
|
|
steveL Watchman
Joined: 13 Sep 2006 Posts: 5153 Location: The Peanut Gallery
|
Posted: Tue Dec 09, 2014 8:08 pm Post subject: |
|
|
Ant P. wrote: | There's the variables-setting facility, runlevel changing, commands to halt/reboot, and also the thing where shutdown/reboot have delays, the delay amount is sent over that pipe. As is any command to cancel a delayed shutdown. |
steveL wrote: | Hmm I'm tempted to say "if it ain't broke." ;)
..
Still, if we have to support the variable setting, it'd be difficult to get rid of /dev/initctl |
Followed up here which seems like a better setting (thanks Shamus.)
Seems like people don't like the "it ain't broke" idea, so let's cleanup this bit and make it play nicely with runit et al as well, before we move on to the more fun stuff, higher up the stack. |
|
Back to top |
|
|
Shamus397 Apprentice
Joined: 03 Apr 2005 Posts: 218 Location: Ur-th
|
Posted: Fri Feb 06, 2015 3:16 pm Post subject: |
|
|
More on the political front: How does systemd relate to Plasma?
Of course all the language is couched in a 'shucks, systemd is technically better for all these reasons and we want to tie KDE to this stuff, so why not' type of language. Guess it won't be long before the Funtoo guys will have to make a shim for KDE as well. |
|
Back to top |
|
|
Ant P. Watchman
Joined: 18 Apr 2009 Posts: 6920
|
Posted: Fri Feb 06, 2015 5:33 pm Post subject: |
|
|
I read that article as a laundry list of things wrong with KDE's design.
Quote: | When people reference standards, someone will cite XKCD, I think more of it like this: (img) |
Which, funnily enough, sounds a lot like GNOME's way of thinking. Which of GNOME or KDE will become the new standard at the expense of the other?
Sometimes to write good software you have to put a little effort into it.
Quote: | Logind allows the session leader to open input devices which are otherwise only available to root.
This allows kwin to monitor key and mouse events whilst blocking other applications from sniffing your keystrokes. |
"X11 allows the session leader to open input devices which are otherwise only available to root..."
"HAL allows the session leader to open input devices which are otherwise only available to root..."
"Consolekit allows the session leader to open input devices which..."
"Logind allows the..."
"$REDHAT_CADT_DECKCHAIR_REARRANGEMENT_OF_2018 allows..."
Quote: | This approach is similar to what we do in our date and time config module; we have our GUI and we also have a small helper app that's DBus activated and runs as root which we are shipping ourselves. Our helper app then runs one of 6 different ntp programs; almost all of which are now outdated.
We don't want to be writing code that runs as root and as a user you really don't want us to be doing so. |
I don't want a DE running system-level services directly! Who the hell thought that was a good idea to code in the first place?! Ubuntu's desktop did NTP setup sanely years ago, why won't KDE? What happened to all that PackageKit bloat *you DE people* invented *for this use case*?
Quote: | Currently in order to achieve the same effect we have a constantly running daemon that monitors /etc/localtime for file changes and then emits DBus signals for all the other apps.
If we could rely on everyone setting the timezone through timedated we can remove this watch which will free some resources for everyone. |
...I'm just speechless at this
Quote: | What else could we make use of? |
A sense of perspective would be a start.
Quote: | User Units allow Plasma services to use the same features as available to system daemons. (...) |
Funnily enough, that's exactly how I make use of runit's tools to start e17 and a bunch of background processes. But my version costs a total of about 100 kilobytes extra RAM; what will systemd do to KDE's already-miserable startup performance? |
|
Back to top |
|
|
Anon-E-moose Watchman
Joined: 23 May 2008 Posts: 6145 Location: Dallas area
|
Posted: Fri Feb 06, 2015 6:00 pm Post subject: |
|
|
Yet more reasons for me to run, not walk, away from kde. :lol _________________ PRIME x570-pro, 3700x, 6.1 zen kernel
gcc 13, profile 17.0 (custom bare multilib), openrc, wayland |
|
Back to top |
|
|
Fitzcarraldo Advocate
Joined: 30 Aug 2008 Posts: 2052 Location: United Kingdom
|
Posted: Fri Feb 06, 2015 6:19 pm Post subject: |
|
|
Shamus397 wrote: | More on the political front: How does systemd relate to Plasma?
Of course all the language is couched in a 'shucks, systemd is technically better for all these reasons and we want to tie KDE to this stuff, so why not' type of language. Guess it won't be long before the Funtoo guys will have to make a shim for KDE as well. |
Thanks for posting the link to that blog post by Dave Edmundson; I hadn't seen it. I'm not surprised, but I am disappointed. I have just left a comment on his post. I've been using KDE (with OpenRC) on my main laptops since 2007 and it's frustrating to learn I would have to stop using something that gives me no trouble. I hope at least one half-decent DE does not force its users to have systemd installed. _________________ Clevo W230SS: amd64, VIDEO_CARDS="intel modesetting nvidia".
Compal NBLB2: ~amd64, xf86-video-ati. Dual boot Win 7 Pro 64-bit.
OpenRC systemd-utils[udev] elogind KDE on both.
My blog |
|
Back to top |
|
|
truekaiser l33t
Joined: 05 Mar 2004 Posts: 804
|
Posted: Sat Feb 07, 2015 4:07 am Post subject: |
|
|
Fitzcarraldo wrote: | Shamus397 wrote: | More on the political front: How does systemd relate to Plasma?
Of course all the language is couched in a 'shucks, systemd is technically better for all these reasons and we want to tie KDE to this stuff, so why not' type of language. Guess it won't be long before the Funtoo guys will have to make a shim for KDE as well. |
Thanks for posting the link to that blog post by Dave Edmundson; I hadn't seen it. I'm not surprised, but I am disappointed. I have just left a comment on his post. I've been using KDE (with OpenRC) on my main laptops since 2007 and it's frustrating to learn I would have to stop using something that gives me no trouble. I hope at least one half-decent DE does not force its users to have systemd installed. |
I stopped using kde when their file indexing program started deleting files out of my home directory on my laptop.. |
|
Back to top |
|
|
mrbassie l33t
Joined: 31 May 2013 Posts: 821 Location: over here
|
Posted: Sun Feb 08, 2015 10:43 am Post subject: |
|
|
I was looking forward to kde5.
Oh well. |
|
Back to top |
|
|
ArneBab Guru
Joined: 24 Jan 2006 Posts: 429 Location: Graben-Neudorf, Germany
|
Posted: Sun Feb 08, 2015 11:06 am Post subject: |
|
|
mrbassie wrote: | I was looking forward to kde5. |
same for me ☹ _________________ Being unpolitical means being political without realizing it. - Arne Babenhauserheide ( http://draketo.de )
pkgcore: So fast that it feels unreal - by doing only what is needed. |
|
Back to top |
|
|
truekaiser l33t
Joined: 05 Mar 2004 Posts: 804
|
|
Back to top |
|
|
Fitzcarraldo Advocate
Joined: 30 Aug 2008 Posts: 2052 Location: United Kingdom
|
Posted: Tue Feb 10, 2015 1:20 pm Post subject: |
|
|
Over the past months it has been somewhat amusing to read a few of the blog posts of Debian user Robert Pogson, as well as the comments on those posts from his "(professionally trained) troll":
To set the scene, in his January 21, 2014 blog post Pogson wrote:
Quote: | So, I did a little test. I have a virtual machine running Debian GNU/Linux “jessie” testing packages. I use it to run the newest version of Gnumeric (because I like the SVG charts/graphs). I apt-get installed everything systemd in Debian’s repository, just 7.5MB of disk space and the damned thing worked immediately after I rebooted. To my surprise booting took half the time of the SYSV method. I’m OK with that. Also, my network mounts and Gnumeric worked exactly as before as did my SSH setup. I have read that there are advantages for security as well and better control by administrators. I’m OK with that. Let them figure out the details…
All in all, I’m leaning to systemd. At least my minimal test worked and caused me no pain. As long as Red Hat doesn’t try to force me to use RPM again, I’m fine with that.
see Lennart Poettering’s blog all about systemd. If he improves GNU/Linux, speeds up booting and doesn’t break my system, he’s my hero. |
What could go wrong? ...
September 2, 2014 Overturning The Distro
October 6, 2014 Just Say No
October 28, 2014 Systemd Debates
November 24, 2014 I GIve Up On Systemd
February 2, 2015 Oh, GAWWWD! Poettering’s Dreams Of Domination Don’t End With systemd
February 9, 2015 systemd Or Poettering, Name Your Poison _________________ Clevo W230SS: amd64, VIDEO_CARDS="intel modesetting nvidia".
Compal NBLB2: ~amd64, xf86-video-ati. Dual boot Win 7 Pro 64-bit.
OpenRC systemd-utils[udev] elogind KDE on both.
My blog |
|
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
|
|