If my assessment isn't acceptable... We can take the upstreams (http://www.paramiko.org/changelog.html) which states:
Warning *This is a backwards incompatible change.* However, *it should only affect installation* requirements; *no API changes are intended or expected*. Please report any such breakages as bugs. Which is exactly as I stated... installation requirement change but no API change. On 6/8/2016 9:02 PM, Jeremy T. Bouse wrote: > I've looked at the paramiko.RSAKey code for 1.16, 1.17 and 2.0 ... Aside > from changing from python-crypto to python-cryptography I see no > evidence to support that the API is incompatible as they have same > exposed calls and parameters. Those functions simply use the different > library calls under the hood. Unless other packages are making calls to > internal functions they shouldn't be, which would be a bug in and of > itself, there shouldn't be any issues. The paramiko package will pull in > python-cryptography instead of python-crypto and there is no Conflicts > in place to hinder their package still maintaining python-crypto in > their dependencies. > > > On 6/8/2016 8:12 PM, Thomas Goirand wrote: >> On 06/08/2016 03:33 PM, Jeremy T. Bouse wrote: >>> Thank you for the report Thomas; however, if you'd taken a beat to check >>> or even contact the maintainer (ie- me) you would have found that this >>> is already being worked on... That said I'm not bothering with >>> experimental as it's a released version of Paramiko. >>> >>> https://github.com/jbouse-debian/paramiko/tree/upstream/2.0.0 >> IMO, you should bother indeed, because the paramiko.RSAKey API is >> incompatible with what was in 1.x, so you should let other maintainers >> fix their packages, and give them enough time to do so (ie: transition). >> >> Cheers, >> >> Thomas Goirand (zigo) >>
