Fix docker swarm mode

[mathcantin]

Fixes #216

The fix is instead of using service.update(image=tag) we need to use service.update(image=f"{tag}@sha256:{latest_image_sha256}").

With a little bit of refractor to remove duplicate code.

[mathcantin]

On the issues, there are this comment

as another way to look at it he is saying it tries to update something that doesnt need updating, which is odd since its checking the sha when it DOES exist initially. Maybe the root can be found there

Still a good idea, but that not cover by this PR.

[dirtycajunrice]

have you ran this code? i like the baseobject but a lot of stuff just wont work. e.g. digest or something doesnt return digest or something. it returns true or false

[mathcantin]

Yes, I try with and without swarm mode and that works. The method _get_digest, return the sha256 of the image, in all cases (dryrun on or off with auth or not). I don't know where you see a method that return a boolean.

[mathcantin]

I understand, the code:

        digest = image.attrs.get("Descriptor", {}).get("digest")
        digest = digest or image.attrs.get("RepoDigests")[0].split('@')[1]
        digest = digest or image.id

        return self._remove_sha_prefix(digest)

is exactly the same thing then that

    digest = image.attrs.get("Descriptor", {}).get("digest")
    if not digest:
            digest = image.attrs.get("RepoDigests")[0].split('@')[1]
    if not digest
        digest = image.id

    return self._remove_sha_prefix(digest)

Because the or keyword isn't a comparator and don't return True or False. I can change it, if you prefer the second solution.

[dirtycajunrice]

Yeah i spoke before i ran the code myself. its completely fine. Im reading over the change right now. Give me a bit its a large change.

[dirtycajunrice]

Awaiting feedback