VSIDO systemd and upstart discussion

[lwfitz]

I have spent a lot of time over the last few days thinking about this discussion while working on Windows/Outlook and cursing profusely.

While I have some experience with systemd, I have none with upstart so I guess my opinion is somewhat uninformed and biased based on functionality on a number of different hardware configurations.
I run both Intel and AMD along with ATI and Nvidia graphics and have never had any real issues with systemd. So I think you can see where my vote goes.

As for the copyright agreement, well, I personally would prefer to see such a large part of this amazing opensource operating system stay opensource.

[jedi]

Wow.  What can I say that statmonkey hasn't?  Not much if anything at all...

At around the 40 minute mark in the first video posted above, when Steve Langasek starts in with the "what's up with Debian and licensing", I definitely threw in the towel on Upstart.  Not that I hadn't already, but when Canonical starts telling dev's to send in their work, and don't bother signing the CLA, and see what happens, that was it for me.  I had read in a couple of different places on the net where some of Ian's (Ian Jackson, former Canonical/Ubuntu dev) posts sounded like he was going to go for Upstart.  I have to say after watching the above, that I don't think you can 'bank' his vote on Upstart after all.  He got pretty intense the second time around with the microphone.

At what price?  At any price?  Just asking.  Having a test version with upstart would be a good thing (if not too much work) but your research has led you to favor systemd.

I'm not sure I would even try Upstart at this point. (it would sure be a hard thing to do)  Steve Langasek so upset me with his cavalier attitude about the licensing issue, that I can't possibly see any way I would ever use it.  The really, really sad part about that is that I know for a fact that sysVinit will be gone soon.  I think I already posted in this thread somewhere that it was an inevitability.  It just has too many flaws and shortcomings for "today's Linux".  Debian has as much as said so.  So, in answer to statmonkey's question about price, I'm not sure there could be a price placed on it.  Debian has been around for coming up on 21 years.  I believe that with this one vote the Tech Committee has in front of them, they could undo everything Debian has done.

I also watched more than one Lennart Poettering video when I finished the one featured above.  He may not have the best social skills in the world, but compared to the two guys (Steve and James) giving the Upstart pitch, he was far and away more informative.  Also, did anyone count how many times they said Ubuntu in their presentation?  I almost thought it was a commercial for Canonical.  (well I guess it was wasn't it?)  Anyway, tho I can't follow all of the technical details they talk about when comparing the two init systems, I know enough to smell a turd on a paper plate, even if it is trying to be hawked as shit on a silver platter.

In reality, by my definition, Upstart really isn't a complete init system in the first place.  To top that off, listening to them trying to answer the questions about some of the functionality and them saying "well we'll try to fix that at some point in the future" (I'm paraphrasing here, watch the video) kind of says a lot.  By comparison, systemd seems to me to already be a valid replacement init system.  Would Arch, or RedHat/Fedora, two of the giants in the realm of Linux, have switched to it already?  Of the over 400 dev's working on systemd, 241 of them came over from udev when they merged them.  Need something patched?  Send it to us!  With Upstart, need something patched?  Sign a copyright agreement and maybe we'll consider it.  As to dev's at Upstart, I know they have at least 3, Scot, James, and Steve.  I've also heard that Canonical limits the amount of dev's on their team and, in the end, has total control over it.

As statmonkey stated, once Canonical has copyrighted code running in Debian, it is the beginning of the end for Debian.  No longer will they be adhering to the FSF, FOSS, GNU, OpenSource philosophy they were founded on.  Not only that, but it will allow for a privately held company to suddenly be making decisions for a very large part of Linux's community.  It will also make it easier over time for them to actually end up "owning" Debian.  So for me this whole thing stinks.  It just does.  I'm actually pretty stunned that Debian has let it get to this point.

I hope Debian is made up of more than just this Tech Committee and has the sense to step back and look at who is voting, and also how they voted.  Freedom right?  I want to know the names of the guys that decided with one vote to destroy what I consider to be the best thing that ever happened to Linux.  (Sorry RMS)  Speaking of him, I can't even begin to imagine his thought process on this whole debacle.  I suppose we'll find out soon enough whether or not there are any ethics and integrity left at Debian.

Last, there has to be an init system.  We know it wont be sysV.  OpenRC, from all I've read, is not in the running for several good reasons.  Upstart is totally not acceptable as an alternative simply because of the CLA/Copyright issues.  (I don't care if the Upstart code is handed down from a deity on golden discs of wonder)  That leaves systemd.  I don't believe, as I once did that this is the lesser of two evils.  I believe it is the only logical choice that a FOSS base could make.  Not just that, but I believe it is the BEST fit and a SOLID choice that will take Debian far into the future.  I also happen to believe that it follows the KISS principle in that it ends up making things simpler in the long run for the folks building distro's.  When you look at systemd in depth, at what it can do, and what it gives in benefits over sysV, then there really is no choice is there?

Yeah, sorry for the novel length post, but as you can probably tell this is getting under my skin just a little bit...

[statmonkey]

Wow.  What can I say that statmonkey hasn't?  Not much if anything at all...

Well for somebody who doesn't have much to say, you sure managed and managed it well.

At around the 40 minute mark in the first video posted above, when Steve Langasek starts in with the "what's up with Debian and licensing", I definitely threw in the towel on Upstart.  Not that I hadn't already, but when Canonical starts telling dev's to send in their work, and don't bother signing the CLA, and see what happens, that was it for me.  I had read in a couple of different places on the net where some of Ian's (Ian Jackson, former Canonical/Ubuntu dev) posts sounded like he was going to go for Upstart.  I have to say after watching the above, that I don't think you can 'bank' his vote on Upstart after all.  He got pretty intense the second time around with the microphone.

Anyway, tho I can't follow all of the technical details they talk about when comparing the two init systems, I know enough to smell a turd on a paper plate, even if it is trying to be hawked as shit on a silver platter.

Wow Jedi, I didn't know you could even get this fired up.  I have always thought of you as the mellow one of the bunch. LOL at the shit on a silver platter.  I love that one.  I agree that Ian seems pretty unlikely to vote for upstart especially after Steve dissed him with the legal committee comment.  Yes, as I said before Lennart at least comes across as a human being.  I think most of his presentation problems have to do with English not being his native language and don't think he realizes how poorly he comes off some times.  I am much more inclined now to give the guy a break on a lot of that.

In reality, by my definition, Upstart really isn't a complete init system in the first place.  To top that off, listening to them trying to answer the questions about some of the functionality and them saying "well we'll try to fix that at some point in the future" (I'm paraphrasing here, watch the video) kind of says a lot.  By comparison, systemd seems to me to already be a valid replacement init system.  ...  I'm actually pretty stunned that Debian has let it get to this point.

All valid points but most of all I agree, why has Debian even let this get this far?  I no longer see anything to consider after just a week of researching this.  Great job pointing out the "we'll fix that later".  There were so many things going on in their presentation it was hard to grok all the obfuscations.

Not just that, but I believe it is the BEST fit and a SOLID choice that will take Debian far into the future.  I also happen to believe that it follows the KISS principle in that it ends up making things simpler in the long run for the folks building distro's.  When you look at systemd in depth, at what it can do, and what it gives in benefits over sysV, then there really is no choice is there?

I also think this point is why Ian, Lennart, etc seem so frustrated.  Also I noticed in the 3 or so videos of Lennart's I watched he acknowledges the elephant in the room openly talking about the differences between upstart and systemd while the upstart guys just dance around and say "well we could tell you why this is so super duper but we don't have enough time".  He also was completely transparent about the dev process and who is involved while they were ... shady.  Finally want to point out that their installed base was all ubuntu customers they had no distro's really to talk about being on board only cloud/ubuntu and LOL wooo weeee the ubuntu phone platform!!!! Gotta get me one of them babies.  If I wasn't nervous about them before I sure as heck am after watching their performance.

[hakerdefo]

Oh Boy. I'm really glad I joined VSIDO community. What a wonderful discussion. Great job @statmonkey, @Digit, @jedi, @PackRat, and of-course the boss @VastOne.
Here is a very detailed comparison of Upstart, OpenRC and systemd projects,

http://www.ohloh.net/p/compare?project_0=Upstart&project_1=OpenRC&project_2=systemd

And I don't have a problem with systemd technically but the people behind systemd are very low on my trust ratings. One is Good Ol' Lennart and he is very popular nowadays so I won't talk about him here. Another key person behind systemd is Kay Sievers. I'll just point to this message in a thread at lkml.org where one Linus Torvalds told Kay Sievers,

Kay, you are so full of sh*t that it's not funny.

Here is the link to the message. You guys should read the whole thread as it will shed some more light on the people who are developing systemd.

https://lkml.org/lkml/2012/10/3/484

Cheers.

[VastOne]

This entire 'choice' is debian bug #727708 and as most of you who have read it, know how scary this is

On the #debian-systemd IRC channel, I have asked why version 208 has not been made as a package for experimental at a minimum.  The answer I have received scares the shit out of me.  For anyone who does not know, we are testing version 204 and updates to 205-208 have significantly made improvements to systemd and squashed some of the naysayers arguments.  It still has bugs, specifically to not cleanly killing network-manager but IMO that should not stop packaging until a decision is made from the committee..

I could build 208 but the principle behind this decision to not package it leads me to believe the writing is already on the wall

[VastOne]

Here is a very detailed comparison of Upstart, OpenRC and systemd projects,

http://www.ohloh.net/p/compare?project_0=Upstart&project_1=OpenRC&project_2=systemd

Cheers.

Wow, the numbers on that comparison are staggering.  The commitment is obvious and the work being done shows at all levels

It's too bad those of us on Debian wanting to see the latest builds cannot since packaging to Experimental has stopped

[statmonkey]

This entire 'choice' is debian bug #727708 and as most of you who have read it, know how scary this is

On the #debian-systemd IRC channel, I have asked why version 208 has not been made as a package for experimental at a minimum.  The answer I have received scares the shit out of me.  For anyone who does not know, we are testing version 204 and updates to 205-208 have significantly made improvements to systemd and squashed some of the naysayers arguments.  It still has bugs, specifically to not cleanly killing network-manager but IMO that should not stop packaging until a decision is made from the committee..

I could build 208 but the principle behind this decision to not package it leads me to believe the writing is already on the wall

Interesting thread, thanks for including it.  I have not seen the statement "we are not going to package until a decision is made"  is that out in the wild?  If I get a moment I may package 208, I am reading now on the network-manager issues.  I had to laugh now one of the Canonical guys is saying that the CLA is no different than the BSD license. HA!

That comparison that hackerdefo posted is likewise informative.  Gosh they are moving a lot of code. Not sure if that is good or bad. 

Also, Colin (I believe) seems to be pushing a dual implementation with both.  Framing it in that light I realize that for me the LICENSE is not the only problem.  I realized that I could not accept that as it would still mean getting into bed with Canonical and having 3 developers in the Canonical Cabal significantly responsible for a major portion of what should be an open/foss system base.  This non-transparency is the issue for me.  I am not sure if there is any level of "significantly better" that upstart could be that would get me over this hump.  There is just no choice. 

In the Linux world we have often had to live with lesser solutions (drivers come to mind) for the freedom that we have.  Yes 'nix is super duper and all that but we have willingly made trade-offs (learning hand coding, cli stuff, accepting that sid will break things) in order to get the flexibility, creative environment, etc. that this world offers us.  If you are telling me to keep my freedom I have to accept that systemd will have some flaws, that it might be slightly less capable than upstart and that I have to put up with the fact that I am not exactly a Lennart/Kay Sievers fan boi and am counting on them.  Sign me up.  It is probably my antidisestablishmentarianist bent but screw - em.  If that means that I have to move to a non-Debian build to get away from having someone (ANYONE) own a major piece of the code and be able to put limitations on my freedom than I will poke my eyes out to spite my face. I know what is going on with systemd.  I know who is doing what and when.  I know where the traps are.  In other words better the devil you know than the one you don't.  I think more and more Canonical is a very worthy devil.  Just my .000000000000000002 cents.

[VastOne]

^ Both you and jedi are really lighting this thread up, well done!

For more arguments from the TC team and what they are going through, check this one out ... Pretty intense

[statmonkey]

[edit] From Ian directly:

So, to recap this and my previous mails and summarise:
* upstart is simpler than systemd (which leads to fewer bugs, etc.)
* upstart integration fits better into a daemon source code
* upstart is easier to package for than systemd
* upstart's community is much better to work with
* systemd's non-portability is (for me) a near-blocker
* upstart's remaining disadvantages are readily fixable SMOP
* upstart is therefore ready for adoption in jessie
* sysvinit has many longstanding bugs and deficiences
* openrc is not ready (I couldn't evaluate it due to lack of a manual)

I therefore conclude that the default init system for jessie should be
upstart.


Thanks,
Ian.

[/edit]
Full post here
https://lists.debian.org/debian-ctte/2013/12/msg00182.html
Is it Game up?

I have to quote one from this exchange https://lists.debian.org/debian-ctte/2013/12/msg00286.html

On Mon, 2013-12-30 at 18:58 +0000, Ian Jackson wrote:
> Also, I get the impression me that the "integration" of much of this
> functionality into the systemd source package has been done for
> political rather than technical reasons.  Indeed to the extent that
> there is a problematically tight technical coupling between these
> components, I find it difficult to avoid doubting the good faith of
> the people making those decisions.  At the very least, I worry that
> the systemd project as a whole is far too willing to impose decisions
> of all kinds on its downstreams.

Your own expressed preference for upstart appeared to be very much
driven by political rather than technical considerations. Using the same
terminology you do, would it not be entirely fair to say that your
decision to support upstart was made in bad faith?

There is comedy, there is slapstick and there is unintentional comedy the gift that keeps on giving.  Two adult males trading this kind of commentary just puts a hop in my step. 

But from the same post a re-iteration of the point I made in my last post.  Finally someone says it ....

> Furthermore, I am much less worried about Debian going it alone
> (although, of course, it's not alone) than you seem to be.  We have
> historically been entirely unafraid to do our own better things, even
> if it is more work and takes us longer.

All interesting stuff.  This is a black hole, for which I really don't have the time.  But it is informative and entertaining.  If it weren't so serious a subject I would probably be on the floor.

[VastOne]

Most polarizing events do capture fancies we never get to see...  The funny thing is, Unity did exactly the same thing and fractured a community to the point of no return

The only difference here, you had a choice with Unity... you could leave Ubuntu/Canonical

Obviously you can do the same with Debian, but like Jedi said earlier that after 21 years of existence, it will be a very painful death

For VSIDO as long as systemd is available it will be used... and I wonder how that would even pan out...

I honestly do not know what I would do if upstart is chosen... (from a personal level) ... I would still build and provide for VSIDO, but would I still use it (debian)?

Probably not in good conscience ...

[statmonkey]

Most polarizing events do capture fancies we never get to see...  The funny thing is, Unity did exactly the same thing and fractured a community to the point of no return

The only difference here, you had a choice with Unity... you could leave Ubuntu/Canonical

Obviously you can do the same with Debian, but like Jedi said earlier that after 21 years of existence, it will be a very painful death

For VSIDO as long as systemd is available it will be used... and I wonder how that would even pan out...

I honestly do not know what I would do if upstart is chosen... (from a personal level) ... I would still build and provide for VSIDO, but would I still use it (debian)?

Probably not in good conscience ...

Very hard to come to terms with this at the moment.  There is a Buddhist precept: Doing wrong without knowing it is wrong is not harmful, doing wrong knowingly, with conscious awareness no matter what the reason creates negative karma. Knowing this precept and not following it is karmic disaster. To paraphrase the Thai version "tam dee dai dee, tam chua dai chua" do good get good, do bad get bad.  In other words I would be done with Debian. The core of why I am here is the open community and contributory nature.  I know that removing that is wrong, I cannot accept wrong. 

How could you, in good conscience provide a solution that violates and uses tools that you don't believe in/are against your core principles?  Not being accusatory it is a real question/quandry that I am struggling with.  I think I need a step back as I come to accept that it looks extremely likely that upstart will be in the next release and in doing so a core of the distribution (even if not used) would be in violation of the principles behind it?

Note that Ian does not address any of these issues in his plus for upstart.  Apparently he got over his license/small dev community concerns.

There religion, philosophy and politics all in one post.  I have now violated all the rules of the forum.  Please take away my hall pass. 

[VastOne]

Like you statmonkey, if upstart is chosen I am walking away from Debian and in doing that, VSIDO would most likely come to it's end

[jedi]

After following these posts here;  Bug#727708: init system other points, and conclusion

I hate to say it, but I don't see Upstart losing.  I do see Debian losing.  I also acknowledge that if Upstart is chosen, I'm through with Debian...

[PackRat]

Like you statmonkey, if upstart is chosen I am walking away from Debian and in doing that, VSIDO would most likely come to it's end

You could base it on Slackware current and use Gslapt/spalt-get for the package management - need to change the name, though.

[zbreaker]

As a still primarily Slackware user I find this of interest. However, I've cut my teeth on Debian based distros, and grown to truly lub VSIDO.
I will keep my thoughts positive and anticipate the best