[08:22] <Keybuk> mbiebl: did your upstart packages never make it through Debian NEW?
[08:23] <mbiebl> Yes, they were rejected because the Conflicted with sysvinit.
[08:23] <Keybuk> interesting...
[08:23] <Keybuk> who rejected them?
[08:23] <Keybuk> Random-J?
[08:23] <mbiebl> jvw
[08:23] <Keybuk> *nods*
[08:23] <mbiebl> It's because sysvinit is essential.
[08:23] <Md> mbiebl: will you use diversions then?
[08:23] <Keybuk> arguably, sysvinit should not be Essential
[08:24] <mbiebl> Md, yes I'll use that for the meantime and target experimental first.
[08:24] <Md> be prepared for a world of pain. m-i-t was not fun...
[08:24] <mbiebl> Keybuk: Yes, but this won't change for etch.
[08:25] <mbiebl> Md: I don't like this idea either.
[08:25] <mbiebl> On the other hand that seems to be the only way to get upstart into the archive now.
[08:28] <Keybuk> mbiebl: you targetted it to experimental, yes?
[08:29] <mbiebl> The one that got rejected? No, it was for unstable.
[08:30] <Keybuk> ahh
[08:30] <Keybuk> if it were experimental, it may have got through
[08:30] <Keybuk> though one of jvw's comments amuses me
[08:31] <mbiebl> Propably.
[08:31] <mbiebl> which one?
[08:31] <Keybuk> - Being not yet an init replacement it's also of limited use yet,
[08:32] <mbiebl> This was a misunderstanding of jvw. He though upstart installed the init binary as /sbin/init and so was not a "drop-in" replacement, yet.
[08:32] <mbiebl> s#/sbin/init#/sbin/upstart#
[08:33] <Keybuk> so he rejected it coming AND going?
[08:33] <Keybuk> he rejected it for not replacing /sbin/init and for replacing it? :p
[08:34] <mbiebl> The main point was, that upstart conflicted with sysvinit, which is essential and that's not allowed by policy.
[08:34] <Keybuk> heh
[08:34] <Keybuk> so is it Debian policy that they will never allow sysvinit to be replaced?
[08:35] <Md> not in etch, and in etch+1 it may still be hard
[08:35] <mbiebl> As long as the essential tag is not removed from sysvinit, yes.
[08:37] <mbiebl> I made a proposal, how we could deal with this by introducing a new base package called "init"
[08:37] <mbiebl> which is essential and depends on "sysvinit | init-system".
[08:37] <mbiebl> This package would be essential and the essential flag from sysvinit would be removed.
[08:38] <mbiebl> Other sysvinit replacements (like upstart, runit etc.) could then add a Conflicts/Replaces/Provides: init-system.
[08:39] <mbiebl> I hope to get that done for etch+1.
[08:39] <Keybuk> the interesting trick will be complying with the invoke-rc.d/update-rc.d policy :)
[08:39] <mbiebl> Yes, the "init" package would provide invoke-rc.d/update-rc.d as mere wrappers.
[08:40] <Keybuk> but how do they work in upstart?
[08:40] <mbiebl> The init systems would install the actual implementation.
[08:40] <Keybuk> "stop 0 23" doesn't mean much
[08:40] <Keybuk> anyway, dinner, bbl
[08:42] <mbiebl> Md, I have asked jvw if he'd accept an upstart package in experimental which does not use diversions but Conflicts/Replaces: sysvinit.
[08:42] <mbiebl> He hasn't answered yet.
[08:42] <mbiebl> Depending on his answer I'll decide to use diversions or not.
[08:42] <Md> cool
[11:35] <Md> aw
[11:35] <Md> oops
[11:45] <maro_> is my setup screwed or is initctl list supposed to look like this? ;) http://borkware.net/~mark/initctl_list.txt
[11:45] <maro_> (0.2.7 looks fine)
[12:07] <mbiebl> maro_: I noticed the same.
[12:07] <mbiebl> Maybe you just file a bug.