[bug#41395] [PATCH 1/2] gnu: python-prompt-toolkit: Update to 3.0.5

[Edouard Klein]

Marius Bakke writes:

> Edouard Klein <address@hidden> writes:
>
>>>>      (synopsis "Interpretable Machine Learning (iML) package")
>>>>      (description "Interpretable ML (iML) is a set of data type objects,
>>>>  visualizations, and interfaces that can be used by any method designed to
>>>> diff --git a/gnu/packages/python-xyz.scm b/gnu/packages/python-xyz.scm
>>>> index b8a51570c3..5fd7e274e9 100644
>>>> --- a/gnu/packages/python-xyz.scm
>>>> +++ b/gnu/packages/python-xyz.scm
>>>> @@ -8922,6 +8922,7 @@ interactive computing.")
>>>>      (build-system python-build-system)
>>>>      (propagated-inputs
>>>>       `(("python-ipykernel" ,python-ipykernel)
>>>> +       ("python-prompt-toolkit" ,python-prompt-toolkit-2)
>>>>         ("python-notebook" ,python-notebook)))
>>>>      (native-inputs
>>>>       `(("python-certifi" ,python-certifi)
>>>> @@ -8950,6 +8951,7 @@ notebooks.")
>>>>      (propagated-inputs
>>>>       `(("python-ipython" ,python-ipython)
>>>>         ("python-traitlets" ,python-traitlets)
>>>> +       ("python-prompt-toolkit" ,python-prompt-toolkit-2)
>>>>         ("python-widgetsnbextension" ,python-widgetsnbextension)))
>>>>      (native-inputs
>>>>       `(("python-nose" ,python-nose)
>>>> @@ -8980,7 +8982,7 @@ in the data.")
>>>>      (propagated-inputs
>>>>       `(("python-ipykernel" ,python-ipykernel)
>>>>         ("python-jupyter-client" ,python-jupyter-client)
>>>> -       ("python-prompt-toolkit" ,python-prompt-toolkit)
>>>> +       ("python-prompt-toolkit" ,python-prompt-toolkit-2)
>>>>         ("python-pygments" ,python-pygments)))
>>>>      (native-inputs
>>>>       `(("python-nose" ,python-nose)))
>>>> @@ -12049,6 +12051,44 @@ collections of data.")
>>>>    (package-with-python2 python-backpack))
>>>
>>> It would be great to update these packages instead of pinning to the old
>>> version.  I won't blame you if you don't feel like embarking on that
>>> journey though.
>>>
>>
>> In the new patches I'm about to send, I updated them. Alas even their
>> latest versions still require python-prompt-toolkit <=2.1, so I had to
>> pin the dependencies. I did it in a cleaner way, though.
>
> Oh too bad that updating did not work.  Terrible that these are so tied
> to the version of python-prompt-toolkit.  Can you split those updates
> out to separate patches, preceding the prompt-toolkit patch?  Then they
> can be reverted and tested individually without having to revert the
> whole thing in case there are problems.
>

Of course :) I should have thought about it, sorry.


>>>> +(define-public python-prompt-toolkit-2
>>>>    (package
>>>>      (name "python-prompt-toolkit")
>>>>      (version "2.0.7")
>>>> @@ -12077,7 +12117,7 @@ collections of data.")
>>>>         ("python-pygments" ,python-pygments)))
>>>>      (native-inputs
>>>>       `(("python-pytest" ,python-pytest)))
>>>> -    (home-page 
>>>> "https://github.com/jonathanslenders/python-prompt-toolkit";)
>>>> +    (home-page "https://github.com/prompt-toolkit/python-prompt-toolkit";)
>>>>      (synopsis "Library for building command line interfaces in Python")
>>>>      (description
>>>>       "Prompt-Toolkit is a library for building interactive command line
>>>> @@ -12104,6 +12144,10 @@ characters, mouse support, and auto suggestions.")
>>>>  (define-public python2-prompt-toolkit-1
>>>>    (package-with-python2 python-prompt-toolkit-1))
>>>>  
>>>> +(define-public prompt-toolkit-2-instead-of-prompt-toolkit
>>>> +  (package-input-rewriting/spec
>>>> +   `(("python-prompt-toolkit" . ,(const python-prompt-toolkit-2)))))
>>>
>>> Is this actually necessary?  Just changing the inputs as you did above
>>> should be sufficient I think.
>>
>> If one dependency is OK with python-prompt-toolkit in version 3, but a
>> package has an implicit dependency on python-prompt-toolkit in version
>> 2, then we either have to pin them all to version 2, and the OK
>> dependency does not get to envoy the update, or we have to resort to this.
>>
>> I did it because python-ipython (19 dependents) is OK with
>> the update, and I did not want to pin it (and all its dependents) to 
>> python-prompt-toolkit-2.
>>
>> This is my first time making such an involved update, so I'm open to
>> suggestions as to alternative ways of doing this.
>
> Oh I see, makes sense.  The new patch is much clearer.  I will send a
> separate reply with further comments.