[ZEPPELIN-1360] Printing long strings is very slow in the python interpreter

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.6.1
• Fix Version/s: 0.7.1, 0.8.0
• Component/s: Interpreters, python-interpreter
• Assignee: Shim HyungSung
• Reporter: Alex Goodman
• Created: 23/Aug/16 18:21
• Updated: 18/Mar/17 16:27
• Resolved: 18/Mar/17 16:27

Description

Printing really long strings is very slow:

s = 's'*100000

This is particularly the case for the python interpreter:
https://puu.sh/qLHA7/0a5f2ac5b3.png

The pyspark interpreter is much better:
https://puu.sh/qLI9x/1ae5cc4e18.png

But this is still very slow compared to running the command in an actual python shell, which takes just a few milliseconds.

This is a very relevant issue for matplotlib integration (ZEPPELIN-1344), which requires us to express raster images as a byte array if we wish to support inline plotting in a notebook. You can easily verify this by using z.show() with the python interpreter. For example, you may try:

%python
import matplotlib.pyplot as plt
import numpy as np
plt.contourf(np.random.random((100, 100)))
z.show(plt)

And notice that this will take a few minutes to run. This is because more complicated plots will require a greater number of characters to represent the image in byte string form.

Issue Links

• Is contained by: ZEPPELIN-1325 Better Python interpreter impl
• is depended upon by: ZEPPELIN-1344 Improving matplotlib integration with zeppelin
• is duplicated by: ZEPPELIN-2071 Very slow parsing of large %html output
• is duplicated by: ZEPPELIN-1894 Matplotlib is very slow in python interpreter
• is part of: ZEPPELIN-1114 Python: improve paragraph ERROR status detection