调试 Python：为什么我的变量不更新？

Question

I'm using PyCharm 2019.2 Professional, Win 10 x64, Python 3.7, and IPython 7.11.1.

When running a script in debug mode and hitting a breakpoint, I can execute statements in the IPython prompt. However, I (sometimes?) cannot change the variables values.

For example, I have a dataframe and check on some condition to get a bool-Series, which I then sum up to check the number of thruths in the dataframe, and if there are none, append another dataframe:

(note that this is a simplified example, I'm not looking to improve this specific snipped)

a_lim = 0.01
my_mask = df_old['A'] < a_lim
if sum(my_mask) == 0:
    df_new = generate_new_df()
    df_old = df_old.append(df_new, ignore_index=True) \
                   .sort_values(by=['A'], ascending=True) \
                   .reset_index(drop=True)
pass

Lets assume df_new contains one row of data that evaluates as True. If I set a breakpoint the if-statement, sum(my_mask) would be 0. If I set my breakpoint at pass , I can check df_old and see the added rows from df_new. At that point, sum(my_mask) is still 0. That is fine.

My Problem:

Stopped at pass , I evaluate my_mask = df_old['A'] < a_lim . Then, I check sum(my_mask) and it still returns 0 . However, if I evaluate sum(df_old['A'] < a_lim) , I will get 1 (the expected result).

What is happening behind the scenes, that Python/IPython seems to selectively update variables, and not others?

Thanks!

Answer 1

I have encountered this bug as well. I have always assumed it is a PyCharm bug not Python. Might be worth raising a bug with JetBrains, think you can do that here:

https://youtrack.jetbrains.com/issues/PY