The following shennanigans were collected during my work on creating a system integration library and framework for hardware and software tests including developer tasks like worktree management, building, target deployment and configuration. Therefore it should be representative for things one might want to do in absence of better tooling. The used Python version was 3.8.2, but most problems still persist.

My tldr; retrospection upfront, which you may be able to reproduce once you try to code long-running services, which recover from all failures, cleanly reset state or generally try to debug spurious problems. Please don't.
Python on itself is great for prototyping up to a few hundred lines of code, like to quickly receive or send some json over tcp/html. However, it is unfeasible to scale, for example to use as library code. Changes in a leaf function can add exceptions to higher level code paths and handling those via exceptions for user friendly error messages, for example to collect context (information along multiple functions, for example from different combination of traversal) becomes unreasonably verbose and error prone. The alternative is to use C-like error handling, which requires to figure out all possible exceptions of Python libstd methods, which language servers do not support (as of 20240404).

Aside of these more fundamental limitations, here the list of shennanigans I have run into:

• xml.dom.minidom breaks space and newlines. Use ElementTree.
• .strip() is necessary after file read, because Python automatically adds \n and there is no way to read without newlines into a list.
• Dictionary is missing this common method, because the implementation is unreadable.

  def is_subdict(small: dict, big: dict) -> bool:
      """
      Test, if 'small' is subdict of 'big'
      Example: big = {'pl' : 'key1': {'key2': 'value2'}}
      Then small = {'pl' : 'key1': {'key2': 'value2'}, 'otherkey'..} matches,
      but small = {'pl' : 'key1': {'key2': 'value2', 'otherkey'..}}
      or small = {'pl' : 'key1': {'key2': {'value2', 'otherkey'..}}} not.
      """
      # since python3.9:
      # return big | small == big
      # also:
      # return {**big, **small} == big
      return dict(big, **small) == big
  

• Dictionary is missing this common method.

  def has_fieldsvals(small: dict, big: dict) -> bool:
    """
    Test, if 'small' has all values of of 'big'
    Example: big = {'pl' : 'key1': {'key2': 'value2'}}
    Then small = {'pl' : 'key1': {'key2': 'value2'}, 'otherkey'..} matches,
    small = {'pl' : 'key1': {'key2': 'value2', 'otherkey'..}} matches,
    and small = {'pl' : 'key1': {'key2': {'value2', 'otherkey'..}}} matches.
    """
    for key, value in small.items():
      if key in big:
        if isinstance(small[key], dict):
          if not has_fieldsvals(small[key], big[key]):
            return False
          else:
            return True
        elif value != big[key]:
          return False
        else:
          return True
      else:
        return False
    return True
  

• Dictionary is missing this common method.

  def merge_dicts(alpha: dict = {}, beta: dict = {}) -> dict:
    """
    Recursive merge dicts. Not multi-threading safe.
    """
    return _merge_dicts_aux(alpha, beta, copy.copy(alpha))
  def _merge_dicts_aux(alpha: dict = {}, beta: dict = {}, result: dict = {}, path: Optional[List[str]] = None) -> dict:
    if path is None:
      path = []
    for key in beta:
      if key not in alpha:
        result[key] = beta[key]
      else:
        if isinstance(alpha[key], dict) and isinstance(beta[key], dict):
          # key value is dict in A and B => merge the dicts
          _merge_dicts_aux(alpha[key], beta[key], result[key], path + [str(key)])
        elif alpha[key] == beta[key]:
          # key value is same in A and B => ignore
          pass
        else:
          # key value differs in A and B => raise error
          err: str = f"Conflict at {'.'.join(path + [str(key)])}"
          raise Exception(err)
    return result
  

• Tuples and dicts are annoying to differentiate

  # dictionary
  dict1 = {
    "m1": "cp",
    "m2": "cp"
  }
  # tuple
  tup1 = {
    "m1": "cp",
    "m2": "cp"
  },
  
  # at least getting the intention correct, but python is still unhelpful with error message
  dict2 = dict({
    "m1": "cp",
    "m2": "cp"
  })
  # tuple
  tup2 = tuple({
    "m1": "cp",
    "m2": "cp"
  }),
  

• Stack trace formatting is inefficient and one can not use gf or gF vim shortcuts to jump to location function to write status + trace to variable.

  def getStackTrace() -> str:
    return repr(traceback.format_stack())
  

• Mixed " and ' strings are invalid json

  # Dict -> str is inconsistent to json -> str, so workaround with
  # dict_asjson_lower = str(dict1).replace("'", '"')
  def combineDictsFromStr() -> None:
    dict1 = {"t1":"val1","t2arr":[{"t2_int":0,"t2_str":"12.0"}],"t3int":30}
    dict1_str_raw = str(dict1)
    dict1_str = dict1_str_raw.replace("'", '"')
    dict2_str = '{"anotherone":"yes", '
    dict2_str +=  '"t3int":30,"t4str":'
    dict2_str += dict1_str + '}'
    dict2 = json.loads(dict2_str)
    _ = dict2
  

• os.kill() does not call registered cleanup function atexit.register(exit_cleanup) by deamonized threads. Must store pids of child processes and clean them explicitly or signal main thread via

  def signalMainThread(self) -> None:
      pass
      # before Python 3.10: _thread.interrupt_main()
      # since Python 3.10: _thread.interrupt_main(signum=signal.SIGKILL)
  

• Socket timeout can cause file-like readline() method to lose data.
  1. Read from Kernel structure and append chunk-wise to buffer from socket until stop event (via select).
  2. After each read, try to line a line from the buffer and remove the line on success (being utf-8 etc).
  3. On failure of reading line, continue with 1.
  4. Teardown should read socket until being empty, if a stop was obtained.
• Generic module annotation is not allowed and mypy has no explicit docs for this. The following does not work and 'module: object' is the closest we can get as simple annotation.

  def check_fn(module: ModuleType) -> int:
      if str(type(module)) != "module": return 1
      return 0
  

• There are no scheduling and watchdog methods, which makes Python thread scheduling very unreliable. Unlucky schedules may cause fatal delay for shuffling data between daemon thread and main thread. As example, an application using 1 main thread and 2 deamon threads may cause the relevant deamon thread not being scheduled for 2 seconds. Empirically 3 seconds work.
• Trailing comma in dictionary or json.dumps generated string has silent failures, for example on parsing the output as json via php.