Hi Russ, Raphael, Marc and dear Lintian maintainers, Le 22.02.2009 05:46, Russ Allbery a écrit : > Raphael Geissert <atomo64@gmail.com> writes: >> Now that lenny is out even oldstable (i.e. etch) has an lsb-base > 3.0-6 >> Am I right to assume that this bug should therefor be closed? > > No, the problem isn't the version of lsb-base, it's that lsb-base isn't > essential and hence can't be assumed to be present without a dependency. > > The solution is probably to make it essential, since it already is > pseudo-essential, but I was holding the bug open until that actually > happens. Right now, you technically have to have the dependency. I think we basically have two options: a) keep lsb-base non-essential; hence add a lintian warning checking that a dependency on lsb-base is present if any of the functions provided by /lib/lsb/init-functions is used (that or checking for the ". /lib/lsb/init-functions" line in the init scripts). (Then, mass-file bugs based on the lintian warning [+ manual check] to ensure that all packages using the functions have the dependency.) b) Mark lsb-base essential; add a lintian warning against all packages depending on lsb-base. (Then, mass-file bugs based on the lintian warning [+ manual check] to ensure that no package has a dependency on lsb-base.) Now, note that there is a bug currently opened against lsb-base (#403120) to have it marked essential (proposing to go the b) route). For now, given that I just recently adopted the lsb source package, I don't want to push for this change, hence marked this bug as wontfix, but I am open to discussion. Finally, I think option a), while creating some more work, is the most safe and future-prone solution (mostly by keeping the essential set as minimum as possible). Cheers, OdyX
Attachment:
signature.asc
Description: OpenPGP digital signature