Code review comment for lp:~denys.duchier/bzr/bzr.ssl

Hi Denys,

Thanks for your patch.

If you have not yet executed our contributor agreement <https://canonical.com/contributors/>, could you please do so?

I would probably prefer bzr+ssl too, especially because we generally have quite explicit url schemes already (bzr+ssh etc).

I agree with John that we should consider the consequences of reusing the same port for encrypted and non-encrypted access. In particular if someone tries to use bzr:// against an ssl port or vice versa, will they get a reasonable message or a confusing one? If it's not reasonably smooth, or plausibly able to be made smooth, then we could either use a different port or use a special command to start encryption.

I suspect if we do add this, the next thing people will want is to do inline authentication over that socket, which probably requires having commands to do so.

This should be mentioned in NEWS.