Cannot define global variables in .pdbrc

[Delengowski]

Bug report

Bug description:

#116660 closed gh-90095 to properly handle commands issued through -c or .pdbrc. The pdb documentation also speaks of convenience variables which is a debugging global variable. I was expecting to be able to define a variable starting with $ as defined in the documentation below, and make manipulations to that variable through commands in a .pdbrc, but doing the following $_capture_list = [] in .pdbrc issues a syntax error

To set a temporary global variable, use a convenience variable. A convenience variable is a variable whose name starts with $. For example, $foo = 1 sets a global variable $foo which you can use in the debugger session. The convenience variables are cleared when the program resumes execution so it’s less likely to interfere with your program compared to using normal variables like foo = 1.

CPython versions tested on:

3.11

Operating systems tested on:

Linux

Linked PRs

• gh-119580: Improve version added section for convenience variable #119583
• [3.13] gh-119580: Improve version added section for convenience variable (GH-119583) #119624
• [3.12] gh-119580: Improve version added section for convenience variable (GH-119583) #119625

[Delengowski]

I may be misunderstanding the paragraph. Perhaps the convenience variable is restricted to the scope of the command and should not be persistent throughout the entire debugging session.

Example program

from itertools import product

x_axis = range(6)
y_axis = range(6)

for x,y in product(x_axis, y_axis):
    ...

In the same directory .pdbrc

$equal_components = []
b 7, (x ==y)
commands
$equal_components.append((x,y))
print($equal_components)
continue

Displays *** SyntaxError: invalid syntax

Instead I attempt

b 7, (x ==y)
commands
$equal_components.append((x,y))
print($equal_components)
continue

Then invoke with python -m pdb example.py and manually type

(Pdb) commands
(com) $equal_components = []
(com) end
(Pdb) continue

Which does not work either, I'm presuming because there is no breakpoint to associate that command with.

What does work is

b 7, (x ==y)
commands
print((x,y))
continue

Breakpoint 1 at /tmp/example_pdbrc.py:7
(Pdb) c
(0, 0)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
(1, 1)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
(2, 2)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
(3, 3)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
(4, 4)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
(5, 5)
> /tmp/example_pdbrc.py(7)<module>()
-> ...
The program finished and will be restarted

[gaogaotiantian]

First of all, you listed that you were using Python 3.11, is that the case? Convenience variable is a feature added in Python 3.12 so it won't work with 3.11.

[Delengowski]

First of all, you listed that you were using Python 3.11, is that the case? Convenience variable is a feature added in Python 3.12 so it won't work with 3.11.

Ah, let me try 3.12 with your backport

[Delengowski]

Defining a convenience variable through .pdbrc does indeed work correctly when using 3.12 with your backport. Although from the paragraph I believe I was misunderstanding how it persists between breakpoints (it doesn't appear to).

With that being said I do find it unclear that convenience variables were added in 3.12 or specific convenience variables ($_frame, etc.) were added in 3.12, or really what of the previous 6 paragraphs was added 3.12.

[gaogaotiantian]

I agree that the added version is not super clear and I can try to improve it. However, about the persistence:

The convenience variables are cleared when the program resumes execution

Your program resumes execution between breakpoints so the convenience variables will be cleared. I think this behavior is well-described.

[Delengowski]

Yes I misunderstood the behavior of the convenience variables. As such I will close this.