Fascinating stuff -- the clearly gifted programmer seemingly came in out of the blue with thousands of lines of changes, which to this day have yet to be adopted into core python (though they do get a shout-out at the top of the official `re` module docs[1] as a "third party module"). Interesting convo pretty much immediately starts up as the haggard maintainers try to respond nicely, and I think one that contains lessons for us all from both sides.
I hadn't realized how popular the `regex` package is (
https://pypistats.org/packages/regex). That's almost as frequently downloaded as `build`, the PyPA standard build front-end, and well ahead of e.g. `ipython`.
One common frustration with Python development is that people with new ideas nowadays are often dismissed "for now" and told to implement their idea as a package and put it on PyPI, and demonstrate that there's demand.
Of course, that puts responsibility for publicizing the idea on the developer. And then if a new package does succeed, suddenly there's no reason why it needs to be brought in to the standard library, where it would become a maintenance burden for core devs, would join "where libraries go to die", etc.
One might imagine, then, that Python has fully embraced the modern idea of having a minimalist standard library while popular third-party packages are installed almost by default, as part of the "ecosystem". A rejection, IOW, of "batteries included". Except - even very conservative changes, like finally removing `distutils` after waiting all the way until 3.12, still nearly cause catastrophe (https://labs.quansight.org/blog/building-scipy-with-flang).
Fascinating stuff -- the clearly gifted programmer seemingly came in out of the blue with thousands of lines of changes, which to this day have yet to be adopted into core python (though they do get a shout-out at the top of the official `re` module docs[1] as a "third party module"). Interesting convo pretty much immediately starts up as the haggard maintainers try to respond nicely, and I think one that contains lessons for us all from both sides.
I hadn't realized how popular the `regex` package is ( https://pypistats.org/packages/regex). That's almost as frequently downloaded as `build`, the PyPA standard build front-end, and well ahead of e.g. `ipython`.
One common frustration with Python development is that people with new ideas nowadays are often dismissed "for now" and told to implement their idea as a package and put it on PyPI, and demonstrate that there's demand.
Of course, that puts responsibility for publicizing the idea on the developer. And then if a new package does succeed, suddenly there's no reason why it needs to be brought in to the standard library, where it would become a maintenance burden for core devs, would join "where libraries go to die", etc.
One might imagine, then, that Python has fully embraced the modern idea of having a minimalist standard library while popular third-party packages are installed almost by default, as part of the "ecosystem". A rejection, IOW, of "batteries included". Except - even very conservative changes, like finally removing `distutils` after waiting all the way until 3.12, still nearly cause catastrophe (https://labs.quansight.org/blog/building-scipy-with-flang).