On Fri, 20 Nov 2020 at 15:44, Ricky Teachey <[email protected]> wrote: > > I was reading the pyinstaller thread and had this idea but didn't want to > hijack. > > Maybe a wild idea, and very possible totally impractical or hopelessly > complex, but: could the existing pypi infrastructure be leveraged into a set > of platform-specific app stores? Maybe maybe we could make it as simple as a > single line in a pyproject.toml file.
It could be done, but I'm not sure what you are thinking of when you say "the existing PyPI infrastructure". Someone would have to write the store code (which maybe could be based on the existing Warehouse code, but should be a separate project). And I wouldn't assume that the donation of space and bandwidth that our existing CDN provides to PyPI would be extended to this as well, but whoever develops this could ask. > Imagine if all the end user does is install the store, and clicks the link to > the project app. The store takes care of the rest. The developer marks their > package as an installable program to be published to the store, optionally > specifying things like specific platforms if needed. All the project website > does for users to install is provide a url to their app in the store. > > Very very rough idea of how it might work: a store app API would provide an > installation GUI if desired, and the store would build the environment needed > for installing the package and dependencies. Once the specified environment > is built, pip would take care of installing the package as usual. > > If this is an awful idea feel free to just say so. I have given it no deep > thought and do not have the expertise to even think through what would be > needed to create such a thing. Like most ideas of this nature, it's pretty good actually - at least in a broad sense. But someone needs to implement it, and everyone currently involved in Python packaging is (a) a volunteer and (b) massively over-stretched. So suggestions like "we need to do X" are unlikely to happen. Suggestions like "I plan on doing X, do I have your blessing" are a different matter (you have mine, FWIW!) :-) Sorry, I hope that doesn't come across as negative. I genuinely don't mean it to. But most of the issues in Python packaging are at their root basically down to no-one having the time - so ideas on their own just add to the backlog. What we really lack is people willing to do the work to implement some of the ideas that are around. And that doesn't really need any sort of approval or support from the PyPA, just (unfortunately) a lot of work and effort... I get that you're saying it's not something you can do yourself. So I guess let's see if anyone wants to take it on :-) Paul _______________________________________________ Python-ideas mailing list -- [email protected] To unsubscribe send an email to [email protected] https://mail.python.org/mailman3/lists/python-ideas.python.org/ Message archived at https://mail.python.org/archives/list/[email protected]/message/EEP7SMAGVFUT5QAUDFTQP3H2I6YURHMA/ Code of Conduct: http://python.org/psf/codeofconduct/
