Somebody give Roland Wolters a cigar, or some non-carcinogenic prize, for his astute observations that 1) Fedora (and, by extension, RHEL) needs to join the club and start looking at a new init system, and 2) that the Ubuntu folks seem to be well underway with Upstart.
Wolters notes that the Fedora folks are discussing alternatives to the existing init process, which is badly dated given all the new devices and services Linux (and other *nix OSes…) have to cope with these days. He also, wisely, says “there are several things I don’t like about these proposals: I have the feeling that all the other attempts were not even really tested…”
I must admit that I cannot even see reasons against Upstart, because it does deliver the features needed: dependency handling, parallel booting, a connection to D-Bus. At least the first two are also handled by InitNG.
I don’t claim to be an expert on the init process. I’ve written about it in the past, but really only from the system administrator and user’s perspective, not the developer’s perspective. However, I’ve been using Upstart in Ubuntu since it was included in Ubuntu by default in the Edgy Eft release. I’ve found no problems as an end user, and I’ve heard no grumblings from other users, either. So, it looks like a reasonable implementation that addresses most, maybe all, of what Fedora is trying to accomplish. So, wtf do we need Fedora folks to waste their time on yet another implementation of init?
This strikes me as part of the “not invented here” syndrome that afflicts many projects. I’d much rather see all the distros put their heads together and fine-tune what Scott James Remnant and other Ubuntu devs have put together than to have two or three different — and likely incompatible — solutions being developed in tandem.
wtf did Ubuntu need upstart? Why did you not scream NIH at them? Maybe Fedora has good reasons. Why don’t you wait till the draft notes disappear and some discussions happen. Why do you assume that everybody will do new efforts for no particular reason.
Read SJR’s article on Linux.com as to why
UpstartUbuntu needed a new system. Maybe Fedora does, but it doesn’t seem so to me.First off, you are right… Somebody get Roland Wolters a beer. What he just did was call Fedora on something that hasn’t been talked about in the open (outside of the mailing lists) for some time, (BTW, minor discussion over a new init system with Fedora is ongoing since before the release of Fedora Core 5) and to be quite honest it is VERY frustrating.
To be honest, this is nice to hear that some things are being done… and there are more complete considerations that are being made.
Second off, and not to sound offensive here, but this is far from the time to start pulling the Ubuntu (Ub) marketing line. Upstart maybe what Fedora needs at the moment, and it’s performance, from what I have read, is very commendable but all that I have heard is use of this Upstart mainly from Ub and those based on Ub. I would like to hear just one distro outside of Ub that uses upstart with as much dependence has have acquired the same results. Maybe Fedora could be the first, but I am not privy on having something so fundamental on another company that isn’t completely transparent about their intentions. If Canonical decides to pull Upstart to a closed development, that would leave a distro like Fedora in a mess trying to replace such a system…. It’s unlikely, but considering how Mark tends to think, it’s not completely out of the question.
Third off, You have a point… one standard on init’s would be much better than many. Hell, with what I just described… it would be quite avoidable if many outside distros where reliant on it. Sort of what happened with X11, XFree86 and X.org… but meh, what do I know… I am just a user.
Duv
I don’t think it’s offensive — it would be nice to see Upstart outside of Ubuntu, to reduce duplication of effort and to provide wider testing. Given the number of people using Ubuntu, I’d have to say that it seems well tested to me, but there’s really no such thing as too much testing.
I can’t speak to how Mark thinks, except to say that I get the impression he is 1) firmly for free software, and 2) a pragmatist. The only app that I am aware of being in closed development from Canonical is Launchpad, and that’s for a sound reason — it needs to be centralized to work well. Having more than one Launchpad doesn’t seem to make a lot of sense.
And, yes, if Upstart were more widely used, it would do a lot towards opening up development beyond just the Ubuntu/Canonical folks. As I understand it, Upstart is currently free software. If it were widely adopted beyond Ubuntu and Mark and/or the Ubuntu team closed it (which I strongly doubt would happen) it would be entirely possible for users to fork it (a la XFree86) and maintain it separate from Ubuntu. I don’t see a threat there.
“, I’d have to say that it seems well tested to me,”
How. Ubuntu upstart still uses sysvinit scripts throughout by default
“Mark thinks, except to say that I get the impression he is 1) firmly for free software”
Why does he so strongly advocate for the proprietary web service called Launchpad and installing proprietary drivers by default in Ubuntu without even informing the user. Doesn’t sound so firm for free software to me. Quite the opposite in fact.
“Maybe Fedora does, but it doesn’t seem so to me.”
What information are you basing your conclusions on? Prelimenary notes? Fedora hasn’t even decided anything. There is just some drafts in the wiki for a proposal. Don’t bother jumping into conclusions.
“Canonical is Launchpad, and that’s for a sound reason — it needs to be centralized to work well. Having more than one Launchpad doesn’t seem to make a lot of sense.”
Standard Mark shuttleworth excuse. Having a proprietary system just because you want one centralized one DOES NOT make sense. It’s a cheap excuse for proprietary vendor lock-in.
“Why does he so strongly advocate for the proprietary web service called Launchpad and installing proprietary drivers by default in Ubuntu without even informing the user. Doesn’t sound so firm for free software to me. Quite the opposite in fact.”
Jef, it’d really help if you slow down and read the *entire* comment before posting, Mmmkay? Thanks.
First off, thank you for answering soundly. Now on to business…
The idea here is that I have yet to hear anyone inside of Ub shouting for outside adoption (Till now, anyway). Sun seems to like to do that when OpenSolaris is mentioned, with things like D-Trace and ZFS. If Canonical wanted something like that done with Upstart, Mark would move on it. As you said and what I have seen of Mr. Shuttleworth is that he is pragmatic when it come down to it. Either this is at the low end of Canonical priorities or Canonical has other plans with Upstart. Personally, I would like to know what those plans are (if any at all).
If anything I am a little surprised that Upstart is seemingly really good and Canonical seems to be sitting on what is normally called a success? Odd isn’t it. Personally, if they start to toe this little project outside of the Ub-scape, whom knows what will happen.
What can I say, would be interesting no?
Jef, please calm down… Mark is not on trial here. I was only proving a point… anyone remembered how git came into being, with BitKeeper mess? That is the mess I was talking about. I sure that is unnerved many (did to me, could have slowed Linux main development to a crawl)… Fedora doesn’t need that kind of mess if you ask me. Few projects, Ub include, do.