Trailing newline in base64 encoded data

[xl0]

I've noticed that Jupyter notebook will append a "\n" to any base64 encoded data in the output of the cell. This does not cause problems, except that VScode-jupyter strips the "\n".

This results in excessively large diffs when Jupyter notebook and vsccde-jupyter are used on the same codebase. I understand that Jupyter was there first, but to me not having the extra "\n" feels like a better solution than having it. Unless I'm missing a good reason for having it, would it be possible to not add this trailing "\n"?

In my understanding the "\n" is introduced by calling b2a_base64 from session.py:json_default:

def json_default(obj):
    [...]
    if isinstance(obj, bytes):
        return b2a_base64(obj).decode('ascii')
    [...]

Since python 3.6 b2a_base64() accepts a newline=True argument.
The description of the function in 3.5 documentation is:

binascii.b2a_base64(data)
Convert binary data to a line of ASCII characters in base64 coding. The return value is the converted line, including a newline char.
The newline is added because the original use case for this function was to feed it a series of 57 byte input lines to get output lines
that conform to the MIME-base64 standard. Otherwise the output conforms to RFC 3548.

Would it be possible to not include that extra newline by passing newline=False to b2a_base64()?

To reptoduce:

from PIL import Image
Image.new("RGB", (10, 10), "green")

results in

   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAAoAAAAKCAIAAAACUFjqAAAAE0lEQVR4nGNkaGDAA5jwSY5caQCnUgCUBZU3vQAAAABJRU5ErkJggg==\n",
      "text/plain": [       "<PIL.Image.Image image mode=RGB size=10x10>"      ]     },
    ...