A Map of Python | Hacker News Reader

A Map of Python

fi-le.net

175 pointsfi-le26 comments

Open articleView on HN

The article also links to this gem, where you can fly through them like a galaxy (WASD controls):

https://anvaka.github.io/pm/#/galaxy/python?cx=-2700&cy=377&...

How can one build a visualisation like this?

And does something like this exist but for the web?

Anvanka provides detailed instructions here[0]. Nowadays I would also consider using Cosmograph[1].

[0]: https://github.com/anvaka/pm?tab=readme-ov-file#your-own-gra...

[1]: https://github.com/cosmograph-org

That's some buttery smooth navigation.

Looks very cool on mobile!

I worked with Zope/Plone for about 10 years in the 2000s. Without reading the story, I loaded the larger version of the graph and looked for a cluster for Plone.

On my third click, I found it.

Then I read the article which actually stated that Plone is one of distinct clusters. Pretty amazing for a 20+ year old technology

My first job was with plone. I hated it. Then I learned the twisted ecosystem and had a moment of crisis.

Those are the devil pits of the python world.

Thank god for django and asyncio.

Python programming Plone was painful.

However, I learned so much from the entire system.

* The CSS of the Plone theme was a Masterpiece. There is a very good reason why Wikipedia used a near direct copy Plone’s CSS for most of the 2000s. Using just a layer of CSS and minor changes to the templates, I could radically re-theme an entire site in a short amount of time.

* Plone enforced semantic HTML and used XHTML. Regardless of what you think of the value of semantics and XHTML, it thought me how to create well structured HTML at a time when the web was full of very broken HTML4

* While programming was painful, Plone’s UX for content managers was first rate. I was invoked in testing Plone, Joomla, Drupal and WordPress. Plone got top marks by a large margin

* Again too marks for Accessibility. In 2005, I built a Plone site for a nonprofit that worked with the blind. I remember users saying they could not believe how easy Plone was to use using the Jaws screen reader

* Multi-lingual sites with workflows for translators. Last year I ran into a translator who used a Plone site I build 20 years ago. They lamented that none of the sites the work on today are as good as that old Plone site.

* etc

Yes it was a good tech from the user perspective.

If you could make it render fast enough, that is.

And twisted did make use cases possible in python that would have been near impossible otherwise.

If you could debug it, that is.

I maintain a project that publishes a SQLite file containing all package metadata, if you don’t want to use BigQuery or the API to do this kind of analysis

https://github.com/pypi-data/pypi-json-data

It would also be interesting to have "transient" dependencies count towards "Used by metric"

E.g., Request is used by more packages than URLlib3, even though requests requires urllib, so in a sense, urllib is used by many more than requests.

They are called transitive dependencies.

Considering how packages ften have optional and unstated dependencies I'd expect an analysis of transitive dependency chains to become very vague, at best distinguishing between packages that are used mostly directly or, like urllib, mostly indirectly.

>> pyyaml

I keep being surprised at the popularity of this dependency. It has a great name for discovery but the API does not easily provide ways to override or customise behaviours.

The continued popularity suggests it’s probably me holding the tool wrong.

I'm curious what sort of behavior you are wanting to change? And, yeah, don't underestimate the effectiveness of a good name.

Nice! I like the look of the resulting map. I've visualized gentoo packages and their dependencies before [1], but the result is no where near as aesthetic as this.

[1] - https://www.thebacklog.net/2011/04/06/high-resolution-depend...

Are there any languages which have an hierarchical repository which is structured so that one can easily browse and find what one needs? Or, arriving where it ought to be, discover that it does not yet exist?

CTAN seems pretty close, but is tightly focused on TeX, and CPAN is for Perl, not Python.

Hackage has a categories system which can be used to browse through all Haskell packages [0]. Unfortunately the categories are unstandardised, so it’s less useful than it could be — though the Flora project exists, which has improved the UI by standardising categories a bit [1].

[0] https://hackage.haskell.org/packages/

[1] https://flora.pm/categories

This could be improved slightly to make it more performant.

Possibly by slowly exposing data to the user?

I would have expected requests to be more prevalent.

Nitpick: It's PyPI (pronounced pie-pee-eye).

I will always pronounce it pie-pie

PyPy is the jit interpreter, though. Thankfully it's usually clear from context.

I truly despise Click and the ways you're supposed to test it

That is not an iteractive map. Google Chrome is the only piece of software which manages to bring my system to a halt, where the mouse then moves an 0.2 FPS.

That map should have used a mapping server, possibly serving vector tiles.

You're right, evidently my programming wasn't quite up to the task here. Good to see though that there's demand, so I'll try and make a better version implementing something like you are describing.

Hacker News sends its best angry nerds (:

Reply on news.ycombinator.com