Deployment

[michalk8]

Before we put this on PyPI/conda-forge, here's a list of TODOs:

• [x] create README (also add small code sample) #2
• [x] create comprehensive installation intructions for Linux/macOS #2
• [x] add tests for stationary_distribution #3
• [x] use k=m in https://github.com/msmdev/pyGPCCA/blob/main/pygpcca/_sorted_schur.py#L358 - correct @msmdev ?
• [x] check for typos in the docs/fix badge links
• [x] add PYPI_PASSWORD repo secret

After this, we can put in on PyPI and afterwards, we can create a PR to conda-forge.

[michalk8]

Ok, in the meantime, I will revert the bump2version commit.

[michalk8]

(the CI should run faster now, since it's cached - when I modified tox.ini, new cache was created).

[michalk8]

Ok, in the meantime, I will revert the bump2version commit.

Done.

[msmdev]

@michalk8 Even the name is __token__ now in PyPI should it still be named PYPI_PASSWORD in GitHub secrets?

[michalk8]

@michalk8 Even the name is token now in PyPI should it still be named PYPI_PASSWORD in GitHub secrets?

Yes. In PyPI, token name should be __token__. The value of the token, should be saved in this repo's secrets with the name PYPI_PASSWORD.

[msmdev]

Ok, then it should be alright now - will push now.

[msmdev]

@michalk8 ok funny, now it fails because of my dummy email... you can insert a real one there: bernhard-reuter@gmx.de

[michalk8]

Done, you can try again.

[michalk8]

@msmdev Congratulations: https://pypi.org/project/pygpcca/1.0.0/ !

[michalk8]

The markdown (\) is not properly rendered, but that can be fixed.

[msmdev]

@michalk8 its fine for now... Cheeers, it is deployed! I owe you a bottle of champagne or some of my self-brewed beer (yes, I am also a home brewer) :-D

[michalk8]

@msmdev cheers! The home-brewed beer sounds really nice!

[msmdev]

@michalk8 if I would like to make some minor style fixes, like fixing the markdown (\), will this require a new version tag and everything?

[michalk8]

For it to be visible on https://pypi.org/project/pygpcca/, yes.

[michalk8]

I've found the recipe for conda (though it was lost):

{% set name = "pyGPCCA" %}
{% set version = "1.0.0" %}

package:
  name: {{ name|lower }}
  version: {{ version }}

source:
  url: https://pypi.io/packages/source/{{ name[0] }}/{{ name }}/{{ name }}-{{ version }}.tar.gz
  sha256: # TODO

build:
  number: 0
  noarch: python
  script: "{{ PYTHON }} -m pip install . --no-deps --ignore-installed -vv"
  skip: True  # [py<36 and not win]

requirements:
  host:
    - pip
    - python
    - setuptools_scm
  run:
    - python
    - numpy >=1.15.0
    - scipy >=1.5.0
    - docrep >=0.3.1
    - mpi4py >=3.0.3
    - petsc >=3.13.0,!=3.14.0
    - slepc >=3.13.0
    - petsc4py >=3.13.0,!=3.14.0
    - slepc4py >=3.13.0

test:
  imports:
    - pygpcca
  commands:
    - pytest -vv
  requires:
    - pytest
    - pytest-mock

about:
  home: https://github.com/msmdev/pygpcca
  license: LGPL-3.0-or-later
  license_family: LGPL
  license_file: LICENSE.txt
  summary: 'Generalized Perron Cluster Cluster Analysis'
  description: |
    # TODO
  doc_url: https://pygpcca.readthedocs.io/en/latest
  dev_url: https://github.com/msmdev/pygpcca

extra:
  recipe-maintainers:
    - msmdev

Could you please fill-in the description? Once the description is done, I will get the sha256 from PyPI and we can open a PR at conda-forge. I will have to double check if the skip condition is right (think we don't need the not win [Windows]).

[msmdev]

@michalk8 cool - I will fill in the description tomorrow :)

[msmdev]

The markdown () is not properly rendered, but that can be fixed.

Couldn't find out how to get it right -.-

[Marius1311]

Amazing guys, well done, and thank you so much for both of your combined efforts! Could we add @michalk8 and my name somewhere on the Pypi page, either under "Acknowledgements" or in the Maintainers section? We both have pypi accounts, mine is https://pypi.org/user/marius1311/ and Mike's is https://pypi.org/user/michalk8/

[msmdev]

Amazing guys, well done, and thank you so much for both of your combined efforts! Could we add @michalk8 and my name somewhere on the Pypi page, either under "Acknowledgements" or in the Maintainers section? We both have pypi accounts, mine is https://pypi.org/user/marius1311/ and Mike's is https://pypi.org/user/michalk8/

@Marius1311 @michalk8 I added a key contributors section in the README.rst. This will be reflected on PyPI after the next release.

[msmdev]

The markdown () is not properly rendered, but that can be fixed.

Couldn't find out how to get it right -.-

@michalk8 any ideas how to get the markdown right on pypi? Simply \n?

[Marius1311]

Amazing guys, well done, and thank you so much for both of your combined efforts! Could we add @michalk8 and my name somewhere on the Pypi page, either under "Acknowledgements" or in the Maintainers section? We both have pypi accounts, mine is https://pypi.org/user/marius1311/ and Mike's is https://pypi.org/user/michalk8/

@Marius1311 @michalk8 I added a key contributors section in the README.rst. This will be reflected on PyPI after the next release.

wonderful, thanks!

[Marius1311]

Small point: I would add a link to the readthedocs to the 'about' section on github - people will be looking for this first.

[msmdev]

Small point: I would add a link to the readthedocs to the 'about' section on github - people will be looking for this first.

@Marius1311 good point - thx!

[michalk8]

@michalk8 any ideas how to get the markdown right on pypi? Simply \n?

I think it would be <br/>, but I would just change here the .replace('|br|', '\\') into .replace('|br|', '\n').

[msmdev]

I've found the recipe for conda (though it was lost):

{% set name = "pyGPCCA" %}
{% set version = "1.0.0" %}

package:
  name: {{ name|lower }}
  version: {{ version }}

source:
  url: https://pypi.io/packages/source/{{ name[0] }}/{{ name }}/{{ name }}-{{ version }}.tar.gz
  sha256: # TODO

build:
  number: 0
  noarch: python
  script: "{{ PYTHON }} -m pip install . --no-deps --ignore-installed -vv"
  skip: True  # [py<36 and not win]

requirements:
  host:
    - pip
    - python
    - setuptools_scm
  run:
    - python
    - numpy >=1.15.0
    - scipy >=1.5.0
    - docrep >=0.3.1
    - mpi4py >=3.0.3
    - petsc >=3.13.0,!=3.14.0
    - slepc >=3.13.0
    - petsc4py >=3.13.0,!=3.14.0
    - slepc4py >=3.13.0

test:
  imports:
    - pygpcca
  commands:
    - pytest -vv
  requires:
    - pytest
    - pytest-mock

about:
  home: https://github.com/msmdev/pygpcca
  license: LGPL-3.0-or-later
  license_family: LGPL
  license_file: LICENSE.txt
  summary: 'Generalized Perron Cluster Cluster Analysis'
  description: |
    # TODO
  doc_url: https://pygpcca.readthedocs.io/en/latest
  dev_url: https://github.com/msmdev/pygpcca

extra:
  recipe-maintainers:
    - msmdev

Could you please fill-in the description? Once the description is done, I will get the sha256 from PyPI and we can open a PR at conda-forge. I will have to double check if the skip condition is right (think we don't need the not win [Windows]).

@michalk8 how about using the description section of the readme as description:

'Generalized Perron Cluster Cluster Analysis program to coarse-grain reversible and non-reversible Markov State Models. Markov State Models (MSM) enable the identification and analysis of metastable states and related kinetics in a very instructive manner. They are widely used, e.g. to model molecular or cellular kinetics. Common state-of-the-art Markov state modeling methods and tools are very well suited to model reversible processes in closed equilibrium systems. However, most are not well suited to deal with non-reversible or even non-autonomous processes of non-equilibrium systems. To overcome this limitation, the Generalized Robust Perron Cluster Cluster Analysis (G-PCCA) was developed. The G-PCCA method implemented in the pyGPCCA program readily handles equilibrium as well as non-equilibrium data by utilizing real Schur vectors instead of eigenvectors. pyGPCCA enables the semiautomatic coarse-graining of transition matrices representing the dynamics of the system under study. Utilizing pyGPCCA, metastable states as well as cyclic kinetics can be identified and modeled. If you use pyGPCCA or parts of it, please cite Reuter 2018 (https://doi.org/10.1021/acs.jctc.8b00079).'

Or is this too long?

[msmdev]

@michalk8 any ideas how to get the markdown right on pypi? Simply \n?

I think it would be <br/>, but I would just change here the .replace('|br|', '\\') into .replace('|br|', '\n').

Alright! Thx @michalk8.

[msmdev]

@michalk8 any ideas how to get the markdown right on pypi? Simply \n?

I think it would be <br/>, but I would just change here the .replace('|br|', '\\') into .replace('|br|', '\n').

Alright! Thx @michalk8.

@michalk8 actually it is currently .replace("|br|", r"\\") so .replace("|br|", r"\n")?

[michalk8]

Or is this too long?

I would keep it short, this is what we have in CellRank - many packages don't have any description (i.e. numpy/tensorflow...). Maybe just keep the first paragraph.

.replace("|br|", r"\n")?

It should be .replace('|br|', '\n'), without the raw strings.

[msmdev]

Ok, will think about something short.

[msmdev]

'pyGPCCA employs Generalized Perron Cluster Cluster Analysis (G-PCCA) to coarse-grain reversible and in particular non-reversible Markov State Models. If you use pyGPCCA or parts of it, please cite Reuter et al. 2018 (https://doi.org/10.1021/acs.jctc.8b00079).'

[michalk8]

Sounds good, will make a PR to conda-later today and ping you in the related thread.

[msmdev]

@michalk8 guess I will make a new release tag now, since the markdown should be fixed now and deploy it to pypi.

[michalk8]

Sounds good, that way we use to correct hash for conda.

[msmdev]

@michalk8 v1.0.1 is deployed :)

[Marius1311]

Thanks for including the contributions @msmdev, it looks very nice!

[msmdev]

Since it the package is successfully deployed on pypi and conda-forge, I'm closing this now. :) Thx, again to @Marius1311 and especially to @michalk8!

[Marius1311]

Amazing work @msmdev and @michalk8 !