Worry about lack of focus
So we've recently seen a big resurgence in activity on improving Python packaging. First off, thats good, hopefully thats why we are all here. That said, I'm becoming worried about a possible lack of focus, and I know I'm not the only one. There have been many ideas floated, and many PEPs either sketched out, reworked, or are stated to be in planning. I think perhaps we should work out some kind of shortlist of what we think can and should be accomplished in the short term and just keep a running list of topics that need energy but are lower priority. This would reduce the chances of hitting the "fix the whole world at once" situation that we have run in to before in this attempt, which often results in burnout and frustration all around. Just to kick things off here are the rough topics I can think of that I've seen discussed recently (ignoring that many of these are dependent on each other): * Including pip with Python 3.4 * Bundling setuptools with pip * Splitting setuptools and pkg_resources * Replacing the executable generation in pip with something new * Working out how to let pip upgrade itself on Windows * Entrypoints in distutils/the stdlib * Executable generation in distlib * Signing/vetting of releases * General improvements to the wheel format * General improvements to package metadata Apologies for anything I have mis-paraphrased or missed, but that is definitely a lot of things to have up in the air. Just want to make sure we can get everything done without anyone going crazy(er) and that we keep sight of whats going on. --Noah
On Jul 18, 2013, at 8:39 PM, Noah Kantrowitz <noah@coderanger.net> wrote:
So we've recently seen a big resurgence in activity on improving Python packaging. First off, thats good, hopefully thats why we are all here. That said, I'm becoming worried about a possible lack of focus, and I know I'm not the only one. There have been many ideas floated, and many PEPs either sketched out, reworked, or are stated to be in planning. I think perhaps we should work out some kind of shortlist of what we think can and should be accomplished in the short term and just keep a running list of topics that need energy but are lower priority. This would reduce the chances of hitting the "fix the whole world at once" situation that we have run in to before in this attempt, which often results in burnout and frustration all around. Just to kick things off here are the rough topics I can think of that I've seen discussed recently (ignoring that many of these are dependent on each other):
* Including pip with Python 3.4 * Bundling setuptools with pip * Splitting setuptools and pkg_resources * Replacing the executable generation in pip with something new * Working out how to let pip upgrade itself on Windows * Entrypoints in distutils/the stdlib * Executable generation in distlib * Signing/vetting of releases * General improvements to the wheel format * General improvements to package metadata
Apologies for anything I have mis-paraphrased or missed, but that is definitely a lot of things to have up in the air. Just want to make sure we can get everything done without anyone going crazy(er) and that we keep sight of whats going on.
--Noah
_______________________________________________ Distutils-SIG maillist - Distutils-SIG@python.org http://mail.python.org/mailman/listinfo/distutils-sig
As my last email said, I completely agree here. We don't just risk burn out but we also risk blessing solutions that haven't been thought through entirely. We also increase the "churn" of packaging making it more difficult for people who *aren't* following along to figure out what they are supposed to do. If we come in and try to advocate them changing huge swathes of their toolchain or how to do things they are going to get frustrated. ----------------- Donald Stufft PGP: 0x6E3CBCE93372DCFA // 7C6B 7C5D 5E2B 6356 A926 F04F 6E3C BCE9 3372 DCFA
On 19 July 2013 13:26, Donald Stufft <donald@stufft.io> wrote:
On Jul 18, 2013, at 8:39 PM, Noah Kantrowitz <noah@coderanger.net> wrote:
So we've recently seen a big resurgence in activity on improving Python packaging. First off, thats good, hopefully thats why we are all here. That said, I'm becoming worried about a possible lack of focus, and I know I'm not the only one. There have been many ideas floated, and many PEPs either sketched out, reworked, or are stated to be in planning. I think perhaps we should work out some kind of shortlist of what we think can and should be accomplished in the short term and just keep a running list of topics that need energy but are lower priority. This would reduce the chances of hitting the "fix the whole world at once" situation that we have run in to before in this attempt, which often results in burnout and frustration all around. Just to kick things off here are the rough topics I can think of that I've seen discussed recently (ignoring that many of these are dependent on each other):
* Including pip with Python 3.4 * Bundling setuptools with pip * Splitting setuptools and pkg_resources * Replacing the executable generation in pip with something new * Working out how to let pip upgrade itself on Windows * Entrypoints in distutils/the stdlib * Executable generation in distlib * Signing/vetting of releases * General improvements to the wheel format * General improvements to package metadata
Apologies for anything I have mis-paraphrased or missed, but that is definitely a lot of things to have up in the air. Just want to make sure we can get everything done without anyone going crazy(er) and that we keep sight of whats going on.
I'm not too worried about whats in progress... I am worried about disruption when we rush things - e.g. the current broken state of setuptools+pip. -Rob -- Robert Collins <rbtcollins@hp.com> Distinguished Technologist HP Cloud Services
On Jul 18, 2013, at 9:47 PM, Robert Collins <robertc@robertcollins.net> wrote:
I'm not too worried about whats in progress...
I am worried about disruption when we rush things - e.g. the current broken state of setuptools+pip.
-Rob
I'd argue that doing too much at once will lead to rushing things and other brokenness. It spreads people out and provides less eyes on each component and less going back and forth because people just don't have the energy to keep track of everything. ----------------- Donald Stufft PGP: 0x6E3CBCE93372DCFA // 7C6B 7C5D 5E2B 6356 A926 F04F 6E3C BCE9 3372 DCFA
On 19 July 2013 10:39, Noah Kantrowitz <noah@coderanger.net> wrote:
So we've recently seen a big resurgence in activity on improving Python packaging. First off, thats good, hopefully thats why we are all here. That said, I'm becoming worried about a possible lack of focus, and I know I'm not the only one.
Indeed, I realised I had a timeline sketched in my head, but had never actually shared it (and many people seem to see metadata 2.0 becoming relevant to end users *far* earlier than I had in mind - I don't see it as becoming relevant until some time in the middle of next year). Posted now, though (I started working on it this morning and just hit send before seeing this thread). Cheers, Nick. -- Nick Coghlan | ncoghlan@gmail.com | Brisbane, Australia
participants (4)
-
Donald Stufft -
Nick Coghlan -
Noah Kantrowitz -
Robert Collins