Bug 11491 - LSB header for the init script
Summary: LSB header for the init script
Status: RESOLVED FIXED
Alias: None
Product: dbus
Classification: Unclassified
Component: core (show other bugs)
Version: unspecified
Hardware: Other All
: medium normal
Assignee: Havoc Pennington
QA Contact: John (J5) Palmieri
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2007-07-06 19:35 UTC by Matthias Clasen
Modified: 2011-01-26 10:07 UTC (History)
2 users (show)

See Also:
i915 platform:
i915 features:


Attachments
patch (745 bytes, patch)
2007-07-06 19:36 UTC, Matthias Clasen
Details | Splinter Review
the other patch (640 bytes, patch)
2007-07-07 07:11 UTC, Matthias Clasen
Details | Splinter Review

Description Matthias Clasen 2007-07-06 19:35:37 UTC
Ubuntu has a patch for this, and I just added one to the Fedora package.
Might be worthwhile to have this upstream.
Comment 1 Matthias Clasen 2007-07-06 19:36:14 UTC
Created attachment 10612 [details] [review]
patch
Comment 2 Havoc Pennington 2007-07-06 21:05:39 UTC
Patch looks fine.

The upstream initscript is Fedora/Red Hat specific afaik (there are separate upstream scripts for other distributions, or could be if people wanted). So patches in the spec file aren't theoretically necessary. I think there was another init script patch in the spec file last time I looked also.
Comment 3 Matthias Clasen 2007-07-07 07:11:50 UTC
Created attachment 10615 [details] [review]
the other patch

yeah, this is the other patch.
Comment 4 Frederic Crozat 2007-11-15 09:40:58 UTC
Would it be possible to change second patch to also start at runlevel 2 (like first patch) ?

We are using almost the same patch on MandrivaLinux
Comment 5 Havoc Pennington 2007-11-15 11:20:52 UTC
Re: second patch, hardcoding /bin doesn't seem right for upstream. My current Fedora package seems to have:

/bin/dbus-cleanup-sockets
/bin/dbus-daemon
/bin/dbus-monitor
/bin/dbus-send
/bin/dbus-uuidgen

which looks to me like the spec file does configure --bindir=/bin ? If so, the patch doesn't seem to be needed.

It would be more correct probably to put only the daemon in /bin and leave the rest of that stuff in /usr/bin, in that case we probably need --daemon-bindir=/bin or something configure arg, then use that for the init script.
Comment 6 Matthias Clasen 2007-11-15 11:43:15 UTC
Yeah, we do have --bindir=/bin so the patch is probably unnecessary. 
Wrt to moving the other binaries to /usr/bin, is that worth the trouble ?
Comment 7 Havoc Pennington 2007-11-15 12:27:38 UTC
I doubt it's worth the trouble, just saying it might be more correct ;-) probably someone will show up in bugzilla wanting to do it eventually... 
Comment 8 John (J5) Palmieri 2008-01-15 14:03:23 UTC
Pushed first patch.  Not sure if we want to change the values for second patch since that only makes sense if the messagebus is in /bin due to some people network mounting their /usr directory.  Also dbus-uuidgen does need to be in /bin (the rest could be moved back to /usr/bin) because when the daemon starts up we call dbus-uuidgen which could be before the /usr directory is mounted.
Comment 9 Matthias Clasen 2008-01-15 14:24:51 UTC
network-mounting /usr is a bit of an 80's concept.
It is really time to stop let this holding us back.
But whatever... 
Comment 10 Simon McVittie 2011-01-26 10:07:30 UTC
The first band of the second patch (changing the chkconfig timing to 22 85) seems to have been done last January, and hard-coding /bin seems to have been rejected, so there's nothing more to do here.

(In reply to comment #2)
> The upstream initscript is Fedora/Red Hat specific afaik (there are separate
> upstream scripts for other distributions, or could be if people wanted).

Debian certainly has its own init script, which uses start-stop-daemon; Ubuntu probably uses the Debian script without modification.


Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.