On 4/18/18 5:58 AM, Joel Cross wrote: > On Tue, 17 Apr 2018, at 6:25 PM, Daniele Nicolodi wrote: >> On 4/17/18 8:57 AM, Joel Cross wrote: >>> Package: wnpp >>> Severity: wishlist >>> Owner: Joel Cross <joel+debb...@kazbak.co.uk> >>> >>> * Package name : python-base58 >>> Version : 0.2.5 >>> Upstream Author : David Keijser <keij...@gmail.com> >>> * URL : https://github.com/keis/base58 >>> * License : MIT >>> Programming Lang: Python >>> Description : base58 encode/decode for Python >>> >>> This package contains the following functions, in a form compatible with >>> that >>> used by the bitcoin network: >>> - b58encode >>> - b58decode >>> - b58encode_check >>> - b58decode_check >> >> Hi Joel, >> >> I had a look at this Python package a while ago, and I was very >> surprised to find that its interface does not match the one of the >> functions in the base64 standard library module in the use of strings vs >> bytes. I had to introduce some wrappers to make it sane in that regard. >> >> I see in the upstream repository some commits that suggest that this has >> been fixed (breaking the API probably). However, nothing that hints >> that a release has been made after that. >> >> It would be nice to have the fixed version in Debian and to avoid API >> breakage, at least once the Python package enters the Debian archive. >> >> Cheers, >> Dan > > Thanks Dan. I assume you're referring to the > https://github.com/keis/base58/issues/15 issue? (Output is string, > not bytes)
Hi Joel, I think so, but I'm not sure that is the only deviation from the base64 API. I can have a look later at the wrapper I wrote to adjust the API, if I still have them, I may have simply rewritten the functions. But it should not be hard to check that `b58encode` and `b58decode` behave the same as their base64 counterparts. Cheers, Dan