CellOutputMimeTypeTracker is very inefficient

[r3m0t]

Environment data

• VS Code version: Version: 1.75.0-insider (user setup)
  Commit: a0562ccdb202a29ac35da81c2b5482e517158f32
  Date: 2023-01-18T08:54:05.566Z
  Electron: 19.1.9
  Chromium: 102.0.5005.194
  Node.js: 16.14.2
  V8: 10.2.154.23-electron.0
  OS: Windows_NT x64 10.0.19044
  Sandboxed: Yes

• Jupyter Extension version (available under the Extensions sidebar): 2022.08.100 (sorry)

• Python Extension version (available under the Extensions sidebar): 2022.20.2

• OS (Windows | Mac | Linux distro) and version: Win

• Python and/or Anaconda version: 3.10

• Type of virtual environment used (N/A | venv | virtualenv | conda | ...): N/A

• Jupyter server running: default local

Steps to reproduce:

[NOTE: Self-contained, minimal reproducing code samples are extremely helpful and will expedite addressing your issue]

I was just debugging microsoft/vscode#168142 and I used this minimal example to examine the case of streaming output:

import time
print('Zi3')
time.sleep(10)
print('Z24ihi')

It looks like the behaviour is very N^2 (calling replaceOutputItems with the full text to append stdout to an existing cell) but I also noticed this really suspicious call stack:

The telemetry is only interested in the cell output_type:

https://github.com/microsoft/vscode-jupyter/blob/2022.08.100/src/kernels/jupyter/jupyterCellOutputMimeTypeTracker.node.ts#L82-L108

However, it's actually running line splitting on the entire output:

https://github.com/microsoft/vscode-jupyter/blob/2022.08.100/src/kernels/execution/helpers.ts#L466-L485

Logs

None

[r3m0t]

Oops. Should have checked main branch ;)