[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: SysV init scripts



Hi,

>>"Kenneth" == Kenneth MacDonald <kenny@ed.ac.uk> writes:

Kenneth> [ New questions at the end of this email :) ]

	You are quite correct, /etc/init.d/xdm does have a variable in
 the beginning of the script that determines whether
 /usr/bin/X11/xdmxdm would run at all. We do need to move al these
 silly little beasties out into a separate, easily manipulatable area
 (I think I like the kernel config file approach; make menuconfig was
 a wonderful innovation).

	We also need a ncer frontend than update-rc.d, which will take
 nore parameters and would be designed for interactive use.

	IMHO, with thees two improvements, we would have solved a lot
 of the tunability problems, without bringing n Linuxconf and
 the like (great program, that, but a overkill, at least for me)
Kenneth> Manoj Srivastava <srivasta@datasync.com> writes:

Manoj>  There is none, by default. *You* are supposed to customize
Manoj>  them for yor machine by adding and deleting links.

Kenneth> Upgrading restores the package's idea of which symbolic links
Kenneth> should be in rc?.d.

	Oops. Umm, I think this needs to be re-thought, maybe in a
 re-implementation of update-rc.d

Kenneth> I suppose my query can be distilled to two questions...

Kenneth> (1) Packages should not decide in their rc.d script whether
Kenneth> to start or not (from existance of a file in /etc or
Kenneth> whatever), but instead rely on symbolic links in rc?.d

	OK.

Kenneth> (2) These symbolic links should be regarded as system
Kenneth> configuration if we're expected to customise them after
Kenneth> installation.  Upgrading to a new package should honour the
Kenneth> existing layout.  Calling update-rc.d package defaults NN
Kenneth> will overwrite a system admin's configuration. 

	Yessss. The current method is that we install all the links
 unconditionally, and postinst sets the variables which will prevent
 the actual invocation at  startup. This may require a Policy
 decision, since we are not very configurable right now. 

	Maybe: we give an additional option to update-rc.d, which is a
 comma separated list of run levels to run in (2 or 3)? postinst then
 may not call update-rc.d unless the user agrees to let the thing run
 (instead of editing init.d/* scripts, do or do not call upate-rc.d)

	If links for the script already exists in the rc?.d
 directories, update-rc.d should do nothing unless called with
 --force? 

	manoj

-- 
 Man weeps to think that he will die so soon; woman, that she was born
 so long ago.  -- H. L. Mencken
Manoj Srivastava               <url:mailto:srivasta@acm.org>
Mobile, Alabama USA            <url:http://www.datasync.com/%7Esrivasta/>


--
TO UNSUBSCRIBE FROM THIS MAILING LIST: e-mail the word "unsubscribe" to
debian-devel-REQUEST@lists.debian.org . Trouble? e-mail to Bruce@Pixar.com


Reply to: