Because the Python Packaging Challenge Supervisor, my work has largely focused on enhancing person expertise on PyPI. Because the Python Packaging ecosystem grows from power to power, one of many initiatives I’ve undertaken is to outline a cohesive technique for Python Packaging.
My Mission
Python Packaging is a various panorama dotted with many instruments that cater to generic and area of interest person necessities. As a neighborhood, one thing that’s poorly understood is the place the neighborhood desires to be in 5 years. Understanding the place we need to be is essential to permit us to establish key objectives for the longer term and the way we are going to attain them, and to make sure we place customers on the coronary heart of the whole lot we do. By defining key deliverables pushed by neighborhood consensus, securing funding turns into simpler as we will reveal that the neighborhood imaginative and prescient will drive innovation and allow higher person expertise and adoption of Packaging instruments.
Consumer Survey
As a way to perceive what finish customers are in search of once they use Packaging instruments, Nicole Harris and I labored with the neighborhood to finalize the survey questions. Nicole developed the ultimate set of survey questions. The survey outcomes are summarized in this doc.
From the survey responses, it was clear that whereas Packaging instruments had been enhancing with time, most customers discovered the panorama too advanced to navigate. Primarily based on the survey responses, the technique dialogue was condensed to those topics-
-
Growth of a single Packaging instrument or a extra unified expertise
-
Higher assist for Packaging customers
-
Phasing out legacy programs
-
Supporting one another locally
-
Encouraging long-term contribution
This publish will summarize the dialogue round a single Packaging instrument. The opposite 4 subjects would be the topic of future posts on Focus on.
One instrument to rule all of them?
An amazing majority of customers advisable a extra unified expertise when utilizing Packaging instruments. The phrase cloud beneath exhibits the frequency of options for the open-ended question- “In the event you may enhance just one space of Python packaging, what wouldn’t it be?”
To take this dialogue additional, I invited maintainers and contributors of any Packaging instrument to reply this query:
“Can we cut back the variety of instruments and convey about some type of unification? Can we do the rest to scale back the complexity?”
After an fascinating and full of life dialogue, we nonetheless should not have a transparent consensus of what an answer ought to appear like, however we do have a method ahead.
Since we’re in the beginning of this dialog, we don’t know what unification means but. As Pradyun Gedam urged, there are numerous choices as to what it may look like-
-
Unification of PyPI/conda fashions
-
Unification of the consumer-facing tooling
-
Unification of the publisher-facing tooling
-
Unification of the workflow setups/tooling
-
Unification/Consistency within the deployment processes
-
Unification/Consistency in “Python” set up/administration expertise
-
Unification of the interface of instruments
Primarily based on these standards, there may be (some) consensus in driving ahead unification of workflow setups/tooling and unification of interface of instruments.
The following main query was, if we do go forward with unification of particular parts, which parts of Packaging would supply the perfect resolution. To supply a unified UI/UX for finish customers, among the options that had been urged had been
-
Reusable libraries
-
Increasing pip to incorporate extra functionalities
-
Recommending present instruments corresponding to hatch, poetry and so on.
This dialogue additionally raised a couple of fascinating issues. Understandably, there may be concern over the effort and time that volunteers have devoted in creating instruments and whether or not all this effort will come to naught if we suggest a selected instrument. Because the dialogue continues, there’s a massive query over the half that PyPA performs and whether or not PyPA could be prepared to take a stand on recommending a selected instrument. One concern that retains developing is that traditionally there was poor communication between PyPA, Packaging instruments and finish customers and if something will change this time.
The way in which ahead
As a way to flesh out the small print of the particular resolution, I like to recommend the next steps:
-
Determine the duties {that a} unified resolution has to perform
-
Determine the easiest way to ship the answer. This might be a brand new instrument, increasing an present instrument or creating requirements/libraries.
-
Submit PEPs to make sure neighborhood governance and consensus
-
Guarantee there may be buy-in from finish customers for proposed resolution(s)
-
Outline improvement roadmaps, milestones, key deliverables and timeline
-
Generate funding to assist improvement
-
Develop and ship
-
Talk, talk, talk
This will likely be an extended and intensive course of. However the effort and time invested on this venture will likely be price it as it would end in innovation and higher person expertise.
We will likely be discussing 4 extra huge questions over the following three months that will likely be used to assist us plan the way forward for Packaging at Python. As we proceed the technique discussions, I invite everybody to take part within the dialogue threads on Focus on.
We’re grateful to Bloomberg for generously offering the funding for this position and making this worthwhile work doable.
