On 27/08/15 21:41, Matthias Klose wrote:
> On 08/27/2015 10:24 PM, Simon McVittie wrote:
>> The libstdc++ transition would be easier to track if its shlibs forced
>> a dependency on libstdc++ (>= 5) even for packages not using any of
>> the new symbols.
>
> I'd like to avoid that.
Is there a reliable way for the transition tracker (and the scripts
setting up mass-rebuilds) to distinguish between gcc-5-built and
gcc-4-built packages, other than this?
I did wonder about some hack like setting the symbol version expression to
libstdc++6 #MINVER# | g++-5-abi
but that does potentially complicate the dependency graph by having apt
try a nonexistent alternative.
>> some packages (such as KDE) are apparently
>> migrating, but actually broken in testing.
>
> please could you name such a package and point out why it is broken?
Sorry, no; that's second-hand information, I don't use KDE (or testing)
myself. I'll try to get concrete information on what is/was broken.
S
