Bug#396454: lintian: Check for LSB logging usage in init.d scripts
"Didier 'OdyX' Raboud" <odyx@debian.org> writes:
> Le 01.11.2006 04:14, Russ Allbery a écrit :
>> I'm not at all convinced that using the LSB logging functions is even a
>> good idea, let alone someting lintian should be diagnosing, until the
>> quality of messages omitted using them is up to what could be done
>> before. Right now, switching to the LSB logging functions results in a
>> significant loss of readability and clarity for failure messages.
> I think we can fairly assume that this situation is now widely different
> than what it was back in 2006: the functions from
> /lib/lsb/init-functions are in wide use all around Debian init scripts.
No, actually, I still feel the same way. Nothing significant has changed
about those functions, and I think it's still very hard to produce a
reasonable warning or failure message while using them directly. That's
the main reason why several of my packages continue to not use the LSB
logging functions, and I'm always reluctant to have Lintian tell people to
do something that I'm not willing to do myself in my own packages.
> Now, I think that lintian could still do one thing rather easily: check
> that only DebianPolicy-compliant functions are used. According to
> lsb-base's README.Debian, there are three LSB-mandated functions that
> are not DebianPolicy-compliant (while still being provided by
> /lib/lsb/init-functions) that packages should not use:
> log_success_msg message
> log_failure_msg message
> log_warning_msg message
> The functions that should be used instead are listed in the same
> README.Debian.
> Checking for uses of those should not need much more than grepping the
> init scripts and wouldn't need the "decent shell script parser"
> currently blocking this bug. IMHO.
I'm okay with this, however.
--
Russ Allbery (rra@debian.org) <http://www.eyrie.org/~eagle/>
Reply to: