mseifert04 / iteration_utilities Goto Github PK

Both share a lot of code and these could also support the eq-parameter from split.

The source tarball available on PyPI includes the test_*.py files, but not conftest.py, helper_cls.py or helper_funcs.py. This means the tests cannot be run from this tarball.

Two optional parameters if pred is used:

• check_for_False (instead of True). Could be named truthy or falsy.
• return_predicate (instead of the item return the pred(item).

With getitem and getitem it's pretty useless now. However it's used in a lot of tests.

I expect this could be speed up using a C-function.

This would avoid unnecessary branching (checking for the presence of the argument) and allow for better optimization, especially if only positional arguments are allowed. Some examples are consume and count_items. I propose to split these into separate functions: consume to advance and consume, and count_items to count_items, count_items + filter, and count_items + filter_eq.

assert(times > 0); did not throw an Exception when times=0 was passed to grouper.

Noticed in #34

This may need to be reworked in the other C functions that use assert.

Generally check if the setup.cfg behaves as the setup.py did before

Hi thanks for the library
I was counting on the index of the list so the code could behave correctly, lastly I used pandas but hope no one face the same bug again
the function drop empty lists instead of keeping them as null which change the index

When installing with pyCharm:
https://pastebin.com/vGCEJNUY

With cmd and powerShell:
https://pastebin.com/5BX7761Y

With pip same error

See astropy/astropy#6658 (comment).

Would be nice if it would link to the source, but probably not so easy because a lot is written as C extension.

The tests for empty iterables were neglected in the memoryleak tests. These should be included!

As possible performance boost one could check the last position first in the bisect-helper function.

Why this may be faster:

• The last item in current is the next one to be yielded. The bisect is called every time the next item of the iterable from which the yielded item came from is to be inserted in current.
• If the iterables are somewhat consecutive the next item may be (with high probability) the next item to be yielded again.
• If there are few iterables the normal bisect and the altered bisect will behave quite similar.
• If there are lots of iterables doing one more comparison before starting the bisection will not slow it down unnecessary.

These assumptions are somewhat debatable, I haven't tested it!

Some of the traverse methods aren't called and thus missing in the coverage. Maybe this can be added by gc.collect or such-like

For github actions

The chained, complement, and constant already support the 3.8 vectorcall tp_call, however these don't:

• flip (#274)
• packed (#278)
• nth
• partial (#281)

For the moment it seemed like too much effort but maybe in the future.

A remined of what to do:

pip install delocate
delocate-listdeps path_to_wheel
delocate-wheel -w target_dir path_to_wheel

Not sure if necessary, but better safe than sorry.

Problem

On aarch64, pip install iteration_utilities builds the wheels from source code and then installs it. It requires the user to have a development environment installed on their system. Also, it takes more time to build the wheels than downloading and extracting the wheels from PyPI.

Resolution

On aarch64, pip install iteration_utilities should download the wheels from PyPI.

@MSeifert04, Please let me know your interest in releasing aarch64 wheels. I can help with this.

Edit

I modified this issue title to request support for Python 3.11 as well as (or instead of) 3.10. I also identify a workaround and an alternative package to iteration_utilities below.

I understand that life gets in the way and projects can easily get pushed to the back-burner and eventually off the stove and into unmaintained status 😅. In fact, your package still works fine as long as users have proper build tools installed. This is usually the case for users on Unix-like machines.

Workaround for Windows users

This can be a tall order for Windows users, who don't have build tools out of the box and must install them. Install Visual Build Tools 2022 (tick the box for C++ build tools and perform the ~7GB install), reboot, and you will be able to successfully pip install iteration-utilities. This will also fix issues with packages which depend on iteration_utilities.

Alternatives

There are a few ways to replace your dependency on iteration_utilities if you wish to generalize your app to be easy to install for Windows and Unix-like users alike:

• Check out how the Copier project replaced (copier-org/copier#889) their dependency on iteration_utilities with something called funcy (PyPI, repo, docs). They specifically needed deep flattening, which was a straight dependency swap. funcy may not be feature-equal with iteration_utilities, but it covers lots of the bases.
• Check out the author of iteration_utilities's recommendations (see refs in the first paragraph) for possible substitutes to functions in iteration-utilities, which may be able to satisfy your dependency.
• Roll your own functions, as in this pure Python deepflatten implementation found in the Copier issue above. https://github.com/wilsonwaters/copier/blob/b72bc87e74542d931369c77f7e27681737812181/copier/tools.py#L154-L169.
• Consider compiling wheels for various architectures for iteration_utilities and contributing them back to this project by getting in contact with the owner.

Original comment

Hello. Thank you for the latest bump to support Python 3.9. Are there plans to support Python 3.10 as well? At least one package (copier-org/copier) would benefit from this support.

Low Priority because there is heapq.merge and sorted

Hi! First of all, thank you for your library.

I tried to install it using the pip command (pip install iteration-utilities), however I had no success. The result can be found below.

I am new to Python, so I don't know exactly if this is something to do with the library or not, but it seems to be a library problem ("note: This is an issue with the package mentioned above, not pip."). Maybe this feedback can be useful for improvements.

× Running setup.py install for iteration-utilities did not run successfully.
  │ exit code: 1
  ╰─> [17 lines of output]
      running install
      C:\Users\Me\AppData\Local\Programs\Python\Python310\lib\site-packages\setuptools\command\install.py:34: SetuptoolsDeprecationWarning: setup.py install is deprecated. Use build and pip and other standards-based tools.
        warnings.warn(
      running build
      running build_py
      creating build
      creating build\lib.win-amd64-cpython-310
      creating build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\_additional_recipes.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\_classes.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\_convenience.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\_recipes.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\_utils.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      copying src\iteration_utilities\__init__.py -> build\lib.win-amd64-cpython-310\iteration_utilities
      running build_ext
      building 'iteration_utilities._iteration_utilities' extension
      error: Microsoft Visual C++ 14.0 or greater is required. Get it with "Microsoft C++ Build Tools": https://visualstudio.microsoft.com/visual-cpp-build-tools/
      [end of output]

  note: This error originates from a subprocess, and is likely not a problem with pip.
error: legacy-install-failure

× Encountered error while trying to install package.
╰─> iteration-utilities

note: This is an issue with the package mentioned above, not pip.
hint: See above for output from the failure.

By the way, I'm using Python 3.10.5 on Windows.

The constants in the _utils.py module shouldn't be exposed in the package.

merge is not yet pickleable.

PyArg_ParseTupleAndKeywords does not increase the refcount of O arguments: https://docs.python.org/3/c-api/arg.html#other-objects

Better check if there is any problem!

A few of the functions could use METH_FASTCALL in Python 3.7+.

To get the signature of a functools instance at runtime could be useful. Even if it probably won't work with inspect.

Some time in the future.

pip reports the following warning on install.

DEPRECATION: iteration-utilities is being installed using the legacy 'setup.py install' method, because it does not have a 'pyproject.toml' and the 'wheel' package is not installed. pip 23.1 will enforce this behaviour change. A possible replacement is to enable the '--use-pep517' option. Discussion can be found at https://github.com/pypa/pip/issues/8559

It should be possible to:

• exclude one (better many) types for return
• include more than one type for return

Call https://docs.python.org/3/c-api/exceptions.html#c.PyErr_Print to print the failure reason

In my script I wrote from iteration_utilities import duplicatess in PyCharm (2019.1.4 Community) and see next error message ("Cannot find reference "duplicates" in init.py")
But code works fine.

All I want to know is it pure PyCharm bug or it can be somehow fixed in code?