just to confirm, all the packages linked against the new libgmp should
follow 'coreutils' example and use 'gmp>=5.0', right?
I just saw that gnunet didn't.
Cheers,
Hilton
03-27-2010, 09:19 PM
Ionut Biru
The libgmp update
On 03/28/2010 12:08 AM, Hilton Medeiros wrote:
Hello,
just to confirm, all the packages linked against the new libgmp should
follow 'coreutils' example and use 'gmp>=5.0', right?
I just saw that gnunet didn't.
Cheers,
Hilton
is not a rule that we have to have this restriction. some maintainers
prefer not to have it, some does.
--
Ionut
03-27-2010, 09:19 PM
Pierre Schmitz
The libgmp update
Am Samstag, 27. März 2010 23:08:08 schrieb Hilton Medeiros:
> Hello,
>
> just to confirm, all the packages linked against the new libgmp should
> follow 'coreutils' example and use 'gmp>=5.0', right?
> I just saw that gnunet didn't.
>
> Cheers,
> Hilton
No, it's the other way round. Using versioned deps is just wrong.
--
Pierre Schmitz, https://users.archlinux.de/~pierre
03-27-2010, 10:21 PM
Allan McRae
The libgmp update
On 28/03/10 08:19, Pierre Schmitz wrote:
Am Samstag, 27. März 2010 23:08:08 schrieb Hilton Medeiros:
Hello,
just to confirm, all the packages linked against the new libgmp should
follow 'coreutils' example and use 'gmp>=5.0', right?
I just saw that gnunet didn't.
Cheers,
Hilton
No, it's the other way round. Using versioned deps is just wrong.
Agreed. Not having it means you can not do "pacman -Sy gnunet" which
would update gmp as a side effect if using a versed dep and cause havok
to the rest of your system. Not having it means only gnunet would be
affected.
03-27-2010, 10:50 PM
Hilton Medeiros
The libgmp update
On Sun, 28 Mar 2010 09:21:51 +1000
Allan McRae <allan@archlinux.org> wrote:
> On 28/03/10 08:19, Pierre Schmitz wrote:
> > Am Samstag, 27. März 2010 23:08:08 schrieb Hilton Medeiros:
> >> Hello,
> >>
> >> just to confirm, all the packages linked against the new libgmp
> >> should follow 'coreutils' example and use 'gmp>=5.0', right?
> >> I just saw that gnunet didn't.
> >>
> >> Cheers,
> >> Hilton
> >
> > No, it's the other way round. Using versioned deps is just wrong.
> >
>
> Agreed. Not having it means you can not do "pacman -Sy gnunet" which
> would update gmp as a side effect if using a versed dep and cause
> havok to the rest of your system. Not having it means only gnunet
> would be affected.
Understood. This just saved me... well thought, thanks.