Julia no longer supported by github backend

[setrar]

Description

For the past few months, I have been using github to display Jupyter Notebooks REPORT.ipynb using the Julia language.
The github backend worked like a charm until yesterday.
In the time being, I put a link to manually render the notebook on REPORT.ipynb on nbviewer.
Since Python Notebooks are working fine, I was wondering if it was just Julia causing the issue.
What would be the solution for Github to allow Julia back again on their platform?

Reproduce

1. Click on REPORT.ipynb
2. Expected behavior REPORT.ipynb on nbviewer
3. Error: Jupyter Cells are no longer rendering Julia source code

[stevengj]

I just noticed the same thing with my notebooks, e.g. this one. In particular, code cells are no longer rendered correctly. This is especially noticeable with multi-line code cells, e.g.

in @setrar's link above — you can see that the line breaks are gone, and everything is smashed together. If you look at the raw .ipynb file (which is just text in JSON format), you can see that the underlying data contains line breaks:

  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "201cdbe8-df2f-4c2f-8f3b-9d92c5d05911",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "freq_offset_est (generic function with 1 method)"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "########################################################\n",
    "# Frequency Offset Estimator Function\n",
    "########################################################\n",
    "function freq_offset_est(sₚₛₛ, hₚₛₛ, Nf, m, 𝑓ₛ)\n",
    "\n",
    "    # Frequency offset estimator\n",
    "    Y = zeros(Float64, length(m));\n",
    "    L = length(hₚₛₛ)\n",
    "    t = 0:(1/𝑓ₛ):((L-1)/𝑓ₛ)\n",
    "\n",
    "    for i = 1:length(m)\n",
    "        # vec() denote the operation of reshaping sequence into a one-dimensional vector\n",
    "        # in Matlab it is the .' (per element transpose)\n",
    "        # (.)ˣ is the conjugate operation\n",
    "        Y[i] += abs2.( ∑( exp.(-2π * j * m[i] .*t ) .* (hₚₛₛ)ˣ .* vec(sₚₛₛ[Nf:(Nf + L -1)])));\n",
    "    end\n",
    "\n",
    "    A𝑓ₒ, 𝑓ₒ = argmax(Y) # Frequency Offset Ampitude, Index and adjusted signal\n",
    "\n",
    "    return m[𝑓ₒ], A𝑓ₒ, 𝑓ₒ, Y # m[𝑓ₒ] Index of the transmitted PSS in Hz\n",
    "end"
   ]
  },

which is why it renders fine on nbviewer.

[stevengj]

However, this is not a Jupyter issue — it should be reported to github, since it's a problem with their website.

[stevengj]

I've created a github support ticket: https://support.github.com/ticket/personal/0/2590265

Probably this issue can be closed here, since there is nothing the Jupyter project can do about it.

[setrar]

I've created a github support ticket: https://support.github.com/ticket/personal/0/2590265

Probably this issue can be closed here, since there is nothing the Jupyter project can do about it.

Hi @stevengj , I saw your reply on Julia Discourse. I am closing this ticket.

Thanks

[setrar]

Issue related to github.com Backend https://support.github.com/ticket/personal/0/2589778