Re: Removing Boost 1.46
On 5 June 2012 at 10:51, Julien Cristau wrote:
| On Mon, Jun 4, 2012 at 21:08:23 -0500, Steve M. Robbins wrote:
|
| > Hi,
| >
| > The output from "dak" for removing source boost1.46 lists a number of
| > broken r-deps, which is to be expected. I had expected that all such
| > packages should be part of the Boost transition tracker [1] but
| > surprisingly, two are missing:
| >
| > gpsshogi: gpsshogi [amd64 i386]
| > libosl: libosl1 [amd64 i386]
| >
| > Can these be added to the transition tracker?
| >
| > [1] http://release.debian.org/transitions/html/boost1.49.html
| >
| >
| > There are additionally three broken build-depends, all of which have
| > trivial fixes that I can NMU, if required. I've added a blocking bug
| > for each against this removal bug. Do you want me to NMU these
| > three prior to removal?
| >
| I think the main blocker for boost1.46 removal from testing right now is
| quantlib. Maybe we should give up on that for arm, mipsen and s390, and
| get the existing out of date rquantlib and quantlib-swig binaries
| removed. Maintainer added to cc.
I would be in favour of that.
Quantlib is an interesting library; these two packages (rquantlib,
quantlib-swig) are its only users and I do not think our users (on other
platforms) are better served by holding this back because of the build errors
on the more esoteric platforms -- the combination of Quantlib and
arm/mips/s390 is just too rare.
So as distasteful as removing packages is, in this case we should.
Dirk
|
| As for the rest that would mean removing the following source packages
| from testing:
| boost1.46 bitcoin kraft gpsshogi libosl rdkit yade cinfony pycuda
| pyopencl
|
| Cheers,
| Julien
--
Dirk Eddelbuettel | edd@debian.org | http://dirk.eddelbuettel.com
Reply to: