BUG- <DataNode>.is_up_to_date raises error when never written before

[FlorianJacta]

Description

The <DataNode>.is_up_to_date method breaks when the Data Node has never been written.

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "c:\Users\jacta\.conda\envs\bar-cutting\lib\site-packages\taipy\core\data\data_node.py", line 528, in is_up_to_date
    and ancestor_node.last_edit_date > self.last_edit_date
TypeError: '>' not supported between instances of 'datetime.datetime' and 'NoneType'

How to reproduce

from taipy.config import Config
import taipy as tp

Config.configure_job_executions(mode="standalone")

# Normal function used by Taipy
def double(nb):
    return nb * 2

# Configuration of Data Nodes
input_cfg = Config.configure_data_node(id="somedata", default_data=21)
output_cfg = Config.configure_data_node(id="result")

# Configuration of tasks
first_task_cfg = Config.configure_task(id="double",
                                       function=double,
                                       input=input_cfg,
                                       output=output_cfg)

# Configuration of scenario
scenario_cfg = Config.configure_scenario(id="my_scenario",
                                         task_configs=[first_task_cfg],
                                         name="my_scenario")

if __name__=="__main__":
    tp.Core().run()
    scenario_1 = tp.create_scenario(scenario_cfg)

    print(scenario_1.result.is_up_to_date)

Expected behavior We should not get an error; the result should be False if the Data Node is invalid.

Possible solution:

Change the is_up_to_date function in taipy/core/data/data_node.py, line 510:

@property
def is_up_to_date(self) -> bool:
    """Indicate if this data node is up-to-date.

    Returns:
        False if a preceding data node has been updated before the selected data node
        or the selected data is invalid.<br/>
        True otherwise.
    """

    if self.is_valid:
        from ..scenario.scenario import Scenario
        from ..taipy import get_parents

        parent_scenarios: Set[Scenario] = get_parents(self)["scenario"]  # type: ignore
        for parent_scenario in parent_scenarios:
            for ancestor_node in nx.ancestors(parent_scenario._build_dag(), self):
                if (
                    isinstance(ancestor_node, DataNode)
                    and ancestor_node.last_edit_date
                    and ancestor_node.last_edit_date > self.last_edit_date
                ):
                    return False
    return False

Runtime environment

• Taipy: 3.1.1

Acceptance Criteria

• [ ] Ensure new code is unit tested, and check code coverage is at least 90%
• [ ] #1200

[jrobinAV]

This should target both develop and 3.1 branches.

[yaten2302]

Hey, @jrobinAV, I'm not really familiar with python, but may I give this issue a try? So, like we've to change return self.is_valid to return False in data_node.py, right?

[FlorianJacta]

More than just that. You have to replace the whole function with the one written above @yaten2302 Is that right? @jrobinAV

[jrobinAV]

@yaten2302 Thank you for helping us on this issue.

Today's code for the method is_up_to_date does not work. The problem is that the code assumes the last_edit_date is populated, meaning the data node has already been written at least once. This is not always true.

The code that @FlorianJacta proposed in the description should solve the issue. So, the main objective of this issue is to:

• Exhibit the problem in a unit test
• Replace the whole method with what has been proposed
• Ensure the new code fixes the exhibited problem
• Ensure all the possible use cases are tested

[yaten2302]

Hey @jrobinAV, thanks for assigning this issue to me 👍 @FlorianJacta I've replaced the whole function data_node.py with the one written above. But, I've a doubt, do we've to create a separate branch or we can work directly on the default branch i.e. develop? Also, could you please guide that how to create tests, as I'm not familiar with that?

[yaten2302]

Description

The <DataNode>.is_up_to_date method breaks when the Data Node has never been written.

Traceback (most recent call last):
  File "<string>", line 1, in <module>
  File "c:\Users\jacta\.conda\envs\bar-cutting\lib\site-packages\taipy\core\data\data_node.py", line 528, in is_up_to_date
    and ancestor_node.last_edit_date > self.last_edit_date
TypeError: '>' not supported between instances of 'datetime.datetime' and 'NoneType'

How to reproduce

from taipy.config import Config
import taipy as tp

Config.configure_job_executions(mode="standalone")

# Normal function used by Taipy
def double(nb):
    return nb * 2

# Configuration of Data Nodes
input_cfg = Config.configure_data_node(id="somedata", default_data=21)
output_cfg = Config.configure_data_node(id="result")

# Configuration of tasks
first_task_cfg = Config.configure_task(id="double",
                                       function=double,
                                       input=input_cfg,
                                       output=output_cfg)

# Configuration of scenario
scenario_cfg = Config.configure_scenario(id="my_scenario",
                                         task_configs=[first_task_cfg],
                                         name="my_scenario")

if __name__=="__main__":
    tp.Core().run()
    scenario_1 = tp.create_scenario(scenario_cfg)

    print(scenario_1.result.is_up_to_date)

Expected behavior We should not get an error; the result should be False if the Data Node is invalid.

Possible solution:

Change the _is_up_todate function in _taipy/core/data/datanode.py, line 510:

@property
def is_up_to_date(self) -> bool:
    """Indicate if this data node is up-to-date.

    Returns:
        False if a preceding data node has been updated before the selected data node
        or the selected data is invalid.<br/>
        True otherwise.
    """

    if self.is_valid:
        from ..scenario.scenario import Scenario
        from ..taipy import get_parents

        parent_scenarios: Set[Scenario] = get_parents(self)["scenario"]  # type: ignore
        for parent_scenario in parent_scenarios:
            for ancestor_node in nx.ancestors(parent_scenario._build_dag(), self):
                if (
                    isinstance(ancestor_node, DataNode)
                    and ancestor_node.last_edit_date
                    and ancestor_node.last_edit_date > self.last_edit_date
                ):
                    return False
    return False

Runtime environment

• Taipy: 3.1.1

Acceptance Criteria

• [ ] Ensure new code is unit tested, and check code coverage is at least 90%
• [ ] #1200

Also, when I was trying to reproduce this issue, it's showing this error in the installation of taipy on running command - pip install taipy

[FlorianJacta]

Try to create an entirely new Python environment. Here is how you contribute to Taipy (doc) You have to create a branch and create a Pull Request that we will review

[yaten2302]

Hey @FlorianJacta, I've created a draft PR, could you please review it and let me know if any changes are required. PR link

[FlorianJacta]

Thank you, the R&D will look into it!