Originally posted by gens
View Post
Announcement
Collapse
No announcement yet.
SystemD/Lennard Pottering hate thread
Collapse
X
-
Originally posted by xeekei View PostNo. Now I think you're confusing "main feature" with "unique feature". Systemd's main feature is very much alike all init systems, except it doesn't use scripts to achieve it. It also happens to have more features beside the main feature.
but think about it like this
i wouldn't describe a person as having kidneys
everybody has kidneys
Comment
-
Originally posted by andyprough View PostIs that all you've got? You are a particularly worthless little troll, aren't you now? Bugger off.
to note the definition from wikipedia
"In Internet slang, a troll (/ˈtroʊl/, /ˈtrɒl/) is a person who sows discord on the Internet by starting arguments or upsetting people,[1] by posting inflammatory,[2] extraneous, or off-topic messages in an online community (such as a forum, chat room, or blog), either accidentally[3][4] or with the deliberate intent of provoking readers into an emotional response[5] or of otherwise disrupting normal on-topic discussion."
maybe you can teach me
Comment
-
Originally posted by gens View Postalso to add
for something like running sshd when it is first time needed there is already a lightweight daemon called inetd
all the things needed for a desktop have to be run anyway
Most init systems do not restart correctly dependent services. Why? That looks useful to me.
When systemd does anything, you just answer by "it's easy to do so with existing tools, or easy to implement", yet those tools are not used or and these easy to implement things are not implemented.
Show us a distribution that starts services in parallels, automatically lazy-starts most of them, put them into cgroups, restart failing services and also restart their dependencies, etc...
Comment
-
Originally posted by gens View Postare you saying other init systems can not start or restart a service ?
Originally posted by gens View Postor find out if it's running ?
Upstart is a bit better in this area, but there's at least one years-old bug opened on Launchpad about situations where daemons can escape its tracking.
Originally posted by gens View Postthat blog is propaganda
Comment
-
Originally posted by erendorn View PostBut then, current systems don't socket activate ssh or Bluetooth daemons. Why is that so?
Most init systems do not restart correctly dependent services. Why? That looks useful to me.
When systemd does anything, you just answer by "it's easy to do so with existing tools, or easy to implement", yet those tools are not used or and these easy to implement things are not implemented.
Show us a distribution that starts services in parallels, automatically lazy-starts most of them, put them into cgroups, restart failing services and also restart their dependencies, etc...
i got inetd installed by default
/etc/services is 2400 lines long, among them ssh (not that ssh port number is a secret)
so making it work is np
i'm honestly thinking about what daemons depend on other daemons so much that they will crash
none will crash if a network fails, since its sockets
idk
funny that you say that since a systemd dev has shown that he doesn't care about real bugs in daemons
there is also monit to monitor, that is way more flexible
thus can make better decisions like do you really want to restart a service that failed (nothing ever fails for no reason)
that that tools are not used in desktop distros i think shows more the lack of need for those kinds of tools
it's not as if they aren't shipped with enterprise level distros, they are
i, for example, never ever had problems with daemons
and i used linux over 5 years ago when many drivers were incomplete and you had to make you own xorg.conf
cgroups yes, are good for security
even in slackware i see cgroups becoming a thing
openrc and upstart both have people working on it,
openrc being the one not paid by any company and still they find time to do it
but parallel boot on a server is not something worth installing the latest fad over
also to add
linux was always about choice
saying that everybody should use the most popular distro is bad
like.. 11% of all kernel commits is from independent people
more then any company tied to a distro
and even having different distros with different goals is good
linux as a hole has gained much from small distros, like knoppix for example
even gentoo
forcing systemd is bad because it limits flexibility
you can do things lennart way or...
well you could do it yourself, but they gonna change the API so your work means nothing
(you can see the talk go from "you can fork/do it" to "you could do it, but it's not advised")
anyway
was nice talking to you
problem i got with talking about systemd is its fanboys
honestly for there to be so many people that don't even know what a OS does when booting and still be wiling to repeat what lennart said in that stupid blog of hes just...
it makes me loose some faith in humanity
i mean really, check out this
i tested jack couple days ago, it uses less cpu then PA
and he wonders why pro audio people like jack...
lennarts blog is all about bashing on existing technologies
he bashed on jack
he bashed on other init systems
he bashed on... well good he didn't decide to make another project but he would probably bash on its competition too
it would even be fine if he was objective and didn't lie about them (saying things like "sysvinit cant set the XDG environment variable" is a pure lie)
i will say what is at fault with sysvinit
i will say what are the deficiencies of alsa
i won't needlessly point out flaws of others thou, since it is a dick move
and everybody that lies just to discredit others are assholes, nothing more nothing less
and yet fanboys keep defending him and the half-shit systems he does
(as proven by comments above, in a anti-systemd thread)
so ye, fuck fanboys
and since there is so many of them for systemd, common sense about it is gone
Comment
-
Originally posted by gens View Posti tested jack couple days ago, it uses less cpu then PA
and he wonders why pro audio people like jack...
lennart keeps repeating "zero copy" but PA is streaming by design (i can't find in its documentation how to do mmaped buffers, that alsa for example supports)
not that zero copy (aka DMA copy) is much better then a normal copy (it can even be worse in some cases)Last edited by gens; 07 March 2014, 09:45 AM.
Comment
Comment