On 8 June 2018 at 13:44, Loren Carvalho <comradeloren@gmail.com> wrote:
Hello distutils-sig!

Apologies if this is the wrong mailing list, Sumanah suggested I write to you fine folks to get some feedback/direction on my pip PR.

Here's the link: https://github.com/pypa/pip/pull/5404

There's a bit of discussion in that PR thread, but also in a couple of Issue threads, most notably https://github.com/pypa/pip/issues/5355 (where I first proposed adding the dist options to install) and https://github.com/pypa/pip/issues/5453 (where some other folks seem to be requesting similar functionality).

Granted there's a lot to ingest, but the discussion in the PR has seemed to stagnate and I desperately would like some direction on how to proceed... or IF to proceed, if it's ultimately decided that this is not something pip maintainers want to do, I totally respect that.

I summarize the status quo in my latest comment on the PR, but to reiterate here: my PR adds `--platform`, `--abi`, `--python-version` and `--implementation` as valid arguments to `install` (formerly these options were only available on `download`). They are only usable when invoking `install` with `--target`, per pfmoore's suggestion. Some other folks have suggested that this limitation (the `--target` one) is unnecessary and that merely using the dist options alone is plenty-explicit about intent. The same folks also would like to see the dist options on the `wheel` subcommand.

I personally have no qualms with their suggestion NOR do i have qualms with pfmoore's suggestion, so I am looking for guidance. Thanks so much in advance, I appreciate your time.

The basic idea seems like a good one to me, and starting out with the `--target` restriction doesn't hurt - it's much easier to relax restrictions like that later than it is to figure out whether or not you've correctly handled all the edge cases that arise without them.

The main automatic implication I would personally expect to see is that specifying any of those options also imply "--only-binary :all:" (rather than requiring it to be passed in separately). That could wait to a follow-up PR though (since the current PR just throws an error in that case).


Nick Coghlan   |   ncoghlan@gmail.com   |   Brisbane, Australia