Bug#730824: [Opinions?] Bug#730824: initscripts: document INIT_VERBOSE in /etc/default/rcS

[Dmitry Bogatov]

control: tags -1 +pending

[2019-08-08 22:24] Thomas Viehweger <patchesThomas.Vie at web.de>
> >> I wondered why some messages don't appear even if I had set VERBOSE=yes
> >> Looking at the code I found the solution (setting INIT_VERBOSE to yes).
> >>
> >> What about adding the following lines (or similar) to /etc/default/rcS:
> >>
> >> # be verbose even if kernel commandline contains "quiet"
> >> #INIT_VERBOSE=yes
> > Yes, this definitely can be done. But I'd ask, whether INIT_VERBOSE
> > variable is actually needed.
> >
> > We have $VERBOSE variable, that defaults to "yes", can be set in
> > /etc/default/rcS and may be modified according to kernel option. And
> > also we have $INIT_VERBOSE, that overrides $VERBOSE independent of
> > kernel option.
> >
> > Isn't it is too much complexity for such simple thing? Why would anyone
> > desire non-verbose boot?
>
> Yes, it is very complex.
>
> I don't want to see regular kernel messages, only irregular. So I use the "quiet" parameter.
>
> While debugging enormous shutdown delays I found the solution with INIT_VERBOSE even
> if it was not documented and only deep buried in /lib/init/vars.sh.
>
> The point for this bug report was to just document this variable.

Thomas, I added proposed lines into /etc/default/rcS in git; this change
will be part of next release.

Collegues, lets talk about future improvements.  Was it good idea to plug
another meaning to already used kernel option "quiet"?

Maybe we could simplify things like this (in ascending-priority order):

 * VERBOSE defaults to yes
 * VERBOSE could be set in /etc/default/rcS
 * VERBOSE could be set by kernel option "initverbose=yes" or "initverbose=no".

and drop INIT_VERBOSE at all (with note in NEWS, of course). Opinions?
-- 
Note, that I send and fetch email in batch, once in a few days.
Please, mention in body of your reply when you add or remove recepients.