i sent pull request before.
this is better one
fix ecdsa key generation for current paramiko master
fix ecdsa key generation
Coverage decreased (-0.02%) when pulling 8455a8a on solarw:ecdsa_key_fix into e811e71 on paramiko:master.
This looks like a significant change (despite its size) - as far as I can tell folks have been using ECDSA support since it was implemented, and this feels like it could break that. Can you provide a concrete test case (real test, or even just an explanation of how to manually test) proving that the code is broken without this change? Thanks!
Yes, this is good. I can't find the generate method on the ECDSA test class anymore. I think this should go in as a bugfix.
@lndbrg can you source the change being made here? Is this something that is breaking with newer ecdsa versions or something? Otherwise it's unclear to me why this isn't breaking code that used to work fine.
@bitprophet hello self, I just looked in detail and the line in question is within ECDSAKey.generate which Paramiko never appears to call itself, and which is also probably not used widely by third parties (given its young age and the infrequency of generating new keys among common Paramiko use cases.)
So yea, I'm on board with this now. Still need to test it out (& probably add tests - when I was grepping I saw tests of the other key types' generate methods but not this one) unless @lndbrg did already.
Changelog re #335