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

Re: ifupdown writes to /etc... a bug?

Goswin Brederlow wrote:
> Well, too late and not too late at all. The patch is send.

I am glad that you have set out to implement this idea.
However, I am not convinced that you are implementing it
exactly the right way.  The /etc/volatile.conf file looks
like overdesign to me.  I don't see the need for the path
to /whatever to be configurable by the user.  Shouldn't we
just decide on a path (e.g., /etc/run) and then move the
run-time state files there by patching programs?  Symlinks
can be used to ease the transition.

Anyway, I've taken a look at your patch.  Here are some

> sysvinit (2.84-3.3) unstable; urgency=low
> [...]
> sysvinit (2.84-3.2) unstable; urgency=low
> [...]
> sysvinit (2.84-3.1) unstable; urgency=low

Put change descriptions in one entry in the changelog,
not three.

>    volatile_mnt="`cat /etc/volatile.conf`"

Doesn't allow for comments in /etc/volatile.conf.

>    if [ "$mnt" = "$volatile_mnt" ]; then
>         # we have /run/
>         have_volatile_mnt=yes

Comment is unneeded and is possibly wrong

>    case "$fs" in
>        /dev/*)
>            # $volatile_mnt is on a real fs so fsck it

Case blocks are usually terminated with ';;'.  I'm not sure
what you mean by a "real" fs.  Could you be more explicit

> # Check the $volatile_mnt file system.

Instead of copying large blocks of code, I suggest you
write a function or two.

> #    is on a ro fs until the remount succeeded. If $volatile_mnt
> #    was found in /etc/fstab mount it and copy skeleton files.

I have read the whole thread and I don't recall reading 
anything that explained why a skeleton directory is
needed.  Presumably it is a kluge for backward compatibility.
If so, then this should be mentioned.

Also mention that you empty $volatile_mnt before filling
it with the contents of /etc/volatile-skel/.

> Things to do to use a read-only / filesystem:
> [...]

The new README file needs some minor editing for English

You introduce the bug report with:

> after the big flamewar about writing to /etc on debian-devel ...

I don't think that this discussion has been a flamewar.
I haven't seen many personal attacks.  It's been a good
discussion of many different possible solutions to a problem


Thomas Hood <jdthood0@yahoo.co.uk>

Reply to: