Latest Round Of Debian Systemd vs. Upstart Voting Ends

[pal666]

However, Upstart is Linux specific too until it actually works with another kernel

we can stop this silly discussion of portability, because even sysvinit is linux-specific and not portable
http://teythoon.cryptobitch.de/posts...-init-systems/

[pal666]

I know, I think the official spelling is bad English so I don't use it. In English proper nouns are capitalized (meaning the first letter is a capital and the rest lowercase) and words and letters aren't jammed together awkwardly. I also don't think acronyms/initialisms should have all letters capital unless people actually say the separate letters. So I write "System D" and "Gnu" and and "I-Mac" and so on. I also refuse to say "OS X". That one's "Mac OS".

do you also think httpd is bad english ? what a stupid excuse

[Chaz]

do you also think httpd is bad english ?

Yes.

what a stupid excuse

I don't think you know what "excuse" means.

[pal666]

Yes.
I don't think you know what "excuse" means.

you failed to show ability to think in general
http://www.thefreedictionary.com/excuse

[c117152]

The Linux specific part is technical, and true, and acknowledged by the authors. It will be Linux specific until someone implements the features needed on another kernel.
However, Upstart is Linux specific too until it actually works with another kernel. The fact it doesn't depend on Linux specific features will not make a working port appear out of nowhere. First finish the BSD port and start a Hurd port, then claim it can be used with them.

I was referring to the language, not the actual point.

But if we're on the subject, neither Upstart nor systemd were ported to other kernels. It's just that Upstart might, possibly, technically be portable since it doesn't rely on cgroups.
More importantly, it doesn't matter since both are released under GPL\LGPL so the BSD camp will never embrace them and instead would rather write their own thing like they did with the rest of their user-land.*

*To be fair, the GNU system guys are also doing their own init: http://www.phoronix.com/forums/showt...588#post395588

[stevenc]

neither Upstart nor systemd were ported to other kernels.

I've seen this repeated a lot here. xnox from Canonical/Ubuntu already began porting Upstart to Debian GNU/kFreeBSD for us, it can build, it runs, though it doesn't know yet how to initialise everything for a normal boot. And it's going to lack some functionality though initially. It is viable though in the jessie timeframe.

OpenRC is already much further along; porting to GNU/kFreeBSD is almost finished, and on GNU/Hurd it's maybe almost on a par with sysvinit (which is only used recently; they had their own custom init until now). OpenRC even worked immediately in GNU/kFreeBSD jails!

As for systemd, I don't think any developers even have an interest in porting systemd to non-Linux. Upstream doesn't, and made it clear wouldn't accept patches for portability. Its design doesn't agree with traditional UNIX-like simplicity at all. And clearly they like to rewrite things all the time so it's a moving target. Even keeping up with its interfaces and unit file syntax would be a chore.

[gens]

Actually, there is. And there are also controversial issues in science, problems where you can't really decide with what you have. Relativity and quantum physics have some points where they don't work too well together, but at the same time they are considered to be "true" (as in being generally accepted in their use contexts) theories, so you usually pick the one that "sounds better" for the context, until you have a breakthrough that "proves" them (this doesn't reeeeeally exist in natural sciences like physics, although it does in general for exact sciences like math and computer science, although some things can't be proved in the latter either). And the "everybody is doing it" is really a big thing in science in general, although that's more correlated to how fucked up the system to measure scientists performance is: you need to have lots of publications in popular journals, and this is far easier to do by making incremental science over the hottest topics at the moment. In my college you can't even teach science if you don't have a high enough publication and quote count (and being quoted is also favored by working in a hot topic).

well... no
everybody knows that quantum theory is (sigma 5) plain wrong
thing is everybody also know it works, at least for what we use it
einstein near the end of he's life said it and started working on other theories

the beauty of complex science is that there are many opinions
like string theory for example, i think it's wrong and that the general unified field theory should be taken as default

problem becomes when we start talking about subatomic particles since, by theory, we can not measure anything about them other then some arbitrary energies


but unlike particle physics, computing is quantifiable (meaning everything can be represented in discrete time units)
so you don't have to invent the 5'th dimension when explaining how your program goes about doing things (like some string theorists do)

recommended big fat book
http://www.amazon.com/Computer-Archi...4&sr=1-1-fkmr0 (i found the forth edition on line at some .edu site)

[gens]

PS
the quantum theory can't explain gravity
so they invented a new particle called "graviton", that probably does not exist

that kind of patchworks reminds me of a certain init system *whistles innocently*

[kigurai]

that kind of patchworks reminds me of a certain init system *whistles innocently*

Yes, redoing init the right way, instead of patching old stuff to do things it was never thought to do is quite obviosuly the right way.

[gens]

Yes, redoing init the right way, instead of patching old stuff to do things it was never thought to do is quite obviosuly the right way.

init has not been patched in... 5 years i think
(patch was minor)

and yes i completely agree with you, we should put roller skates on cows, they will get around faster and make better milk
obviously