sysvinit=2.93-2 and upcoming 2.93-3

[Dmitry Bogatov]

Hi!

Given lack of objections to my previus email, I made upload out
`afc866dc' as sysvinit=2.93-2.

Now, I see following agenda on next release:


 * #910289 -- logsave issue. Since transfer of ownership (see bug) is going
   to take some time, I propose to solve issue temporary with calling
   `logsave' only if is availaible.

 * #917139 -- block devices under /run. For now I prepared patch, that
   add required special cases into /etc/init.d/umountfs. Not the most
   principled solution, but given date of release and severity of issue,
   I believe 80% solution is warranted here.

 * #915051 -- obsolete conffile /etc/init.d/motd

Dear co-maintainers, you are welcome to take a look at commits, that
address issues above at branch wip/kaction/into-2.93-3.

Next comes issues, that I would like to address, but no patch prepared
yet. Suggestions welcome.

 * Why bin:initscripts is arch:any? It does not seems to contain any
   arch-dependent files (and given its name, should not). Related
   question -- shouldn't /lib/init/* be /usr/share/init/*?

 * What is the purpose of directory /var/lib/initscripts?

 * #132542 -- move of /etc/init.d/{rc,rcS} out of /etc. After reading
   fhs-3.0, bundled with Policy, I am still not sure, whether it should go
   into /usr/share/sysvinit or /usr/libexec/sysvinit.

 * Why /etc/rc.local is created in so elaborate way, and is not just a
   conffile?

 * Are there anybody around who knows, what is the state of code in
   initscripts.postinst, related to Hurd (lines 167-191). It was last
   touched at 2009; it is still needed?

 * As Debian does not support jump-over-release upgrades, all checks for
   upgrades since 2.88dsf-59 (oldstable) are redundant. Correct?

By the way, explicit stating that you do not have objections to proposed
changes is welcome. I want to make sure that changes are more-or-less
agreed upon, not just overlooked, while trying to address issues in
timely fashion.