[DRE-maint] Bug#991015: Acknowledgement (cannot execute due to incorrect shebang line)

[Adrian Bunk]

On Fri, Jul 16, 2021 at 11:11:44AM -0300, Antonio Terceiro wrote:
> On Tue, Jul 13, 2021 at 12:48:49PM +0300, Adrian Bunk wrote:
> > On Mon, Jul 12, 2021 at 09:10:15PM -0400, Ryan Kavanagh wrote:
> > > This bug seems to be, in part, due to a potentially (?) broken ruby
> > > upgrade behaviour. I've been running unstable on this laptop for ~6
> > > years, but still had
> > > 
> > > ii  ruby2.1 [ruby-interpreter]  2.1.5-4
> > > ii  ruby2.2 [ruby-interpreter]  2.2.4-1
> > > 
> > > installed as my only ruby interpreters. These are no longer available in
> > > unstable, and ruby2.1 was last available in:
> > > 
> > > ruby2.1    | 2.1.5-2+deb8u3 | oldoldstable | source, amd64, armel, armhf, i386
> > > 
> > > Will users upgrading from buster to bullseye will encounter a similar
> > > issue if they've dist-upgraded from jessie to stretch to buster?
> >
> > The problem is the following in the dependencies:
> >   Depends: ruby | ruby-interpreter, ...
> > 
> > The /usr/bin/ruby symlink is in the ruby package,
> > the alternative dependency ruby-interpreter is
> > therefore wrong.
> 
> We noticed that a while ago, and ruby2.2 was the last of the rubyX.Y
> packages to actually provide `ruby-interpreter`.

Thanks for this information.

> At this point, there are 900+ packages that still depend on `ruby |
> ruby-interpreter`, so this isn't something we can easily fix. I will
> submit a lintian check for this to get eventually fixed, but it will
> take a while.
>...

~ 90% of these seem to be Ruby modules, it is not obvious to me that
"ruby | ruby-interpreter" is even a bug for these.

A bookworm MBF for the small part that are non-interpreter packages 
might be an option?

cu
Adrian