Re: Bug#896667: transition: r-base-3.5
On 31 May 2018 at 16:15, Emilio Pozuelo Monfort wrote:
| Hi,
|
| On 31/05/18 15:45, Emilio Pozuelo Monfort wrote:
| > On 31/05/18 15:09, Dirk Eddelbuettel wrote:
| >>
| >> Emilio, Seb,
| >>
| >> Can you confirm that now that we have
| >> a) "green light" on the transition, and
| >> b) the r-base package is in unstable
| >> we should see binary: any packages being rebuilt -- which I do not yet
| >> see. When will this start?
| >
| > I will start the rebuilds soon (i.e. later today).
|
| Scheduled now (it will take some time as it's 320 arch:any packages).
|
| By the way there was a problem with my suggested jdk change: the architecture
| restriction is applied first, and then the first alternative is taken, so for
| e.g. m68k, openjdk-10-jdk is taken as it's the first valid alternative for that
| architecture. But we don't want jdk there at all. So there are two good options:
|
| default-jdk [!arm !hppa !kfreebsd-i386 !kfreebsd-amd64 !hurd-i386]
There must be something that makes this form not preferred as I had been
using the "concretePackage | virtualPackage" form for many years.
If we did this, I would not have to jump through hoops updating the package ...
|
| i.e. drop the openjdk-10-jdk alternative, or
|
| default-jdk [!arm !hppa !kfreebsd-i386 !kfreebsd-amd64 !hurd-i386] |
| openjdk-10-jdk [!arm !hppa !kfreebsd-i386 !kfreebsd-amd64 !hurd-i386]
... yet the other day I needed the form with openjdk-10 (as r-cran-rjava
failed with with given that "its" r-base has still used openjdk-9.
So I think the second form is better. I can do a quick rebuild if you concur.
|
| i.e. add the architecture restriction to openjdk-10-jdk as well.
|
| I'd go with the former, but both should work. If you can apply one of those
| changes that'd help.
|
| Sorry for not realising that earlier.
No worries. It's just cycles :)
Dirk
--
http://dirk.eddelbuettel.com | @eddelbuettel | edd@debian.org
Reply to: