Upstream: Installation broken on MacOS with Apple Silicon

[onedrawingperday]

Describe the bug

• Installing with pip fails as described in Apple Silicon support lutzroeder/netron#627
• Installing from source is possible but the dependencies numpy, scipy and scikit-learn cannot be built on Apple Silicon, therefore the library cannot run.

Current workaround as described in lutzroeder/netron#627 (comment) is to install the above dependencies with a Miniforge installer.

---

Relevant upstream issues:

numpy/numpy#18143
scipy/scipy#13409
scikit-learn/scikit-learn#19137

[SekouDiaoNlp]

Hi @onedrawingperday , thanks for letting me know about this issue.
As I do not have a Mac computer with the M1 processor, I was not aware of this compilation problem.

I will add mlconjug3 to conda forge during the week, so that users with the M1 processor will be able to install mlconjug without any hassle as all libraries are already pre-compiled on conda.

I will let you know as soon as mlconjug3 is available in anaconda/miniconda.

Cheers,

Sekou.

[SekouDiaoNlp]

Hi @onedrawingperday .

I did the necessary to upload mlconjug3 to conda-forge.

It should take a couple of days before it is available for download.

You should be able to install mlconjug3 directly from conda with pre-compiled dependencies once it is on the conda-forge channel.

I will keep you informed as soon as it is released.

Cheers,

Sekou.

[SekouDiaoNlp]

Hello @onedrawingperday !

You can find the pull request for the integration in conda-forge here: issue 14559 on the conda-forge repository

[SekouDiaoNlp]

Hi @onedrawingperday how are you today?

I can finally confirm that mlconjug3 is now available on conda-forge.

You can find the website at conda mlconjug3.

To install it from conda, type the following commands:

conda install -c conda-forge mlconjug3

Would it be possible for you to test that everything works by installing mlconjug3 on a Mac computer with an M1 processor and let me know if everything works accordingly before I close this Pull Request?

Thank you again for letting me know about this issue, I would not have discovered it by myself as I don't have a Mac with anM1 processor handy to make tests.

Cheers,

@SekouDiaoNlp .

[onedrawingperday]

Hi @SekouDiaoNlp

I will check it out during the weekend and let you know.

Thanks!

[SekouDiaoNlp]

HI @onedrawingperday. would it be possible for you to both try mlconjug3 version 3.7.15 and 3.7.16 please?

I just want to make sure that the latest 3.7.16 installs correctly on Mac computers with an M1 processor.

Thanks in advance.

@SekouDiaoNlp

[SekouDiaoNlp]

HI @onedrawingperday.

I just wanted to let you know that I just released the latest version of mlconjug3 which is version 3.7.17.

As I designed a new Continuous Integration/ Continuous Delivery pipeline using GitHub Actions Workflows, I am now able to automatically build, install and test my package on Windows, Linux and MacOS.

Moreover, after uploading and maintaining mlconjug3 on conda-forge, they run every new release through a thorough CI/CD pipeline on Azure DevOps building and testing the package not only on different Operating Systems and Python versions, but also on a large array of different processor architectures.

I am pretty confident that if you conda install -c conda-forge mlconjug3, you will install the latest version 3.7.17 and everything should build and install properly even on a Computer with an Apple ARM M1 processor.

You don't need to test if versions 3.7.15 and 3.7.16 build and install properly.

If you can just confirm me that 3.7.17 builds, install and works fine on a MacBook with an M1 processor, I will close this Pull Request.

Thanks again for letting me know about this issue and for being willing to test the newest version to confirm it works as expected.

Cheers man,

@SekouDiaoNlp

[onedrawingperday]

Hello @SekouDiaoNlp

I just installed miniconda and setup the conda-forge channel.

The latest mlconjug3-3.17.7 builds fine!

Thank you so very much! 👍✌️👏

Click to see the conjugation of manger

mlconjug3 manger
{
"manger": {
"Infinitif": {
"Infinitif Présent": "manger"
},
"Indicatif": {
"Présent": {
"1s": "mange",
"2s": "manges",
"3s": "mange",
"1p": "mangeons",
"2p": "mangez",
"3p": "mangent"
},
"Imparfait": {
"1s": "mangeais",
"2s": "mangeais",
"3s": "mangeait",
"1p": "mangions",
"2p": "mangiez",
"3p": "mangeaient"
},
"Futur": {
"1s": "mangerai",
"2s": "mangeras",
"3s": "mangera",
"1p": "mangerons",
"2p": "mangerez",
"3p": "mangeront"
},
"Passé Simple": {
"1s": "mangeai",
"2s": "mangeas",
"3s": "mangea",
"1p": "mangeâmes",
"2p": "mangeâtes",
"3p": "mangèrent"
}
},
"Conditionnel": {
"Présent": {
"1s": "mangerais",
"2s": "mangerais",
"3s": "mangerait",
"1p": "mangerions",
"2p": "mangeriez",
"3p": "mangeraient"
}
},
"Subjonctif": {
"Présent": {
"1s": "mange",
"2s": "manges",
"3s": "mange",
"1p": "mangions",
"2p": "mangiez",
"3p": "mangent"
},
"Imparfait": {
"1s": "mangeasse",
"2s": "mangeasses",
"3s": "mangeât",
"1p": "mangeassions",
"2p": "mangeassiez",
"3p": "mangeassent"
}
},
"Imperatif": {
"Imperatif Présent": {
"2s": "mange",
"1p": "mangeons",
"2p": "mangez"
}
},
"Participe": {
"Participe Présent": "mangeant",
"Participe Passé": {
"ms": "mangé",
"mp": "mangés",
"fs": "mangée",
"fp": "mangées"
}
}
}
}

[SekouDiaoNlp]

Hi @onedrawingperday, I hope you had a great weekend.

Thank you very much for having taken the time to test that the newest version of mlconjug3 available on conda-forge works flawlessly on MacBooks with an ARM M1 processor.

And thank you again for letting me know about this issue in the first place. I really appreciate it!

I also updated the installation instructions in the documentation to make clear that mlconjug3 can be installed both through:

$ pip install mlconjug3 or

$ conda config --add channels conda-forge
$ conda config --set channel_priority strict
$ conda install mlconjug3

I also added an extra warning note advising users with an ARM M1 computer to install mlconjug3 through conda.

Thank you for your contribution to this project, and do not hesitate to propose changes or enhancement to this package.

By the way, I am interested in the use case you have for mlconjug3, do you use it as a standalone package to conjugate verbs, or do you use it as a library in your own software?

Peace, love and have fun 😍

@SekouDiaoNlp

[onedrawingperday]

@SekouDiaoNlp

Again thank you very much for fixing this! 🙏

My use case is pretty simple:

J'apprends le français 😃

[SekouDiaoNlp]

@SekouDiaoNlp , alors bonne chance dans ton apprentissage.

La meilleure façon d'apprendre le français est de pratiquer la langue avec des locuteurs natifs.
(I am French by the way if it was not obvious).

I can advise you to try to have face to face conversations with french speaking people on Omegle,
your comprehension and fluency will improve dramatically if you try to do it a few times a week.

Good luck in your learning and enjoy your life (even though the lockdown due this fu**ing coronavirus is really starting to pissing me off!

Γάμα τον κοροναϊό 😝

@SekouDiaoNlp