C-Stack usage is too close to limit - not due to recursive code

Question

I have an intruiging question, and I can not seem to find the answer. I have a specific python function, which runs a sql script, gets the data and transforms it into a pd.DataFrame. The size of a typical response is around 100Kb (83.9Kb in the example below):

When I run the exact same function, which I import in an R session using reticulate, I get a "C Stack usage too close to the limit error".

I do not understand why R gives me such a response for such a small dataframe. Does it have to do with the translation of Python code to R via reticulate? Quite a specific question, but I am curious to know if somebody knows the answer

''

Please see the code of the Python function below:

def get_raw_data_(make, year):

    sql_file = WORKING_DIRECTORY.joinpath("sql/get_raw_data.sql")
    sql_string = sql_script_to_string(sql_file)

    dragend_rules_file = WORKING_DIRECTORY.joinpath(f'sql/dragend_rules/{make}.sql')
    dragend_rules =  sql_script_to_string(dragend_rules_file)

    tellend_rules_file = WORKING_DIRECTORY.joinpath(f'sql/tellend_rules/{make}.sql')
    tellend_rules =  sql_script_to_string(tellend_rules_file)

    to_replace = {
        'INSERT_YEAR':year,
        'INSERT_DRAGEND_RULES':dragend_rules,
        'INSERT_TELLEND_RULES':tellend_rules
    }
    
    for key, value in to_replace.items():
        sql_string = sql_string.replace(key,value)

    sql_string = sql_string.replace('INSERT_MAKE', make)
        
    res = execute_sql(sql_string, cur)
    res = pd.DataFrame(res, columns = [
        "CONTRACTNR", "MAKE", "TYPE", 
        "VEHICLEKIND", "STATUS", "LICENSEPLATE",
        "FUELTYPE", "COMMERCIALUNITLEVEL1", "YEARMONTHREGISTRATION",
        "DRAGEND"
    ])

    res = res.reset_index(drop=True)
    cols_fact = res.columns[~res.columns.isin(['CONTRACTNR', 'YEARMONTHREGISTRATION'])]
    res[cols_fact] = res[cols_fact].astype('category')

    return res

I call this function in R as follows:

source('get_raw_data.py')
get_raw_data_('ME', '2021')

Answer 1

I had the same problem, but querying GCP Firestore documents.

After so many hours looking for documentation and trying different options, I managed to solve it (still not sure what was causing the problem) using "reticulate dev version" as said in https://github.com/rstudio/reticulate/issues/885.