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

Re: Bug#747628: perl: module deprecations / removals in 5.20



-=| Russ Allbery, 19.05.2014 13:21:41 -0700 |=-
> Damyan Ivanov <dmn@debian.org> writes:
> 
> > The note in the Release Notes would mean that users of unstable will not
> > notice the removal, but I guess this is a price to be paid.  Bugging
> > each and every upgrade with a NEWS.Debian entry seems like
> > spam. 'perl-base' is ranked #1 in popcon. Hmm, perhaps add a NEWS.Debian
> > entry and remove it before Jessie? (this starts to have too many moving
> > parts :))
> 
> perl-base isn't the relevant package, though, isn't it?  This would be for
> perl.  (Or were some of those modules in base?)

perl has a 99.59% installation rate :)

> I think NEWS.Debian is appropriate.  It's certainly used for things 
> that people generally care about less (such as all new CAs in the
> ca-certificates package), and people have to make an intentional choice to
> install apt-listchanges to see the upgrade notifications.  I think this is
> exactly the sort of situation that NEWS.Debian was intended for: a
> backwards-incompatible change in the packaging.

Thanks. This convinces me. I guess I erred on the "don't possibly harm 
anyone" side.

> I vote for documenting this in NEWS.Debian, retaining Depends 
> *maybe* for one release at most, and then downgrading to at least 
> Recommends with a goal of removing the dependency entirely in the 
> long run.

Niko commented on IRC that removal of automatically installed packages 
that are no longer depended on is fine, since apt/aptitude gives 
a warning about this, IOW there is no silent breakage. And if people 
blindly click "Next>>" it's not something we can fix.

Another point that Niko raised and with whis I tend to agree is that 
Recommends may be better than a hard dependency since that would 
satisfy the needs of the people that care more about disk size than 
about safety


Cheers,
    dam

Attachment: signature.asc
Description: Digital signature


Reply to: