| msg297608 - (view) |
Author: Malcolm Smith (Malcolm Smith) |
Date: 2017-07-03 19:14 |
| https://docs.python.org/3/reference/simple_stmts.html#the-import-statement defers the full specification of relative imports to PEP 328. PEP 328 gives the example "from ...sys import path" in a second-level package, and notes "while that last case is legal, it is certainly discouraged". However, in the current implementation it actually isn't legal. Using a stdlib second-level package to test: Python 3.5.3 (default, Apr 10 2017, 07:53:16) [GCC 6.3.0 64 bit (AMD64)] on win32 Type "help", "copyright", "credits" or "license" for more information. >>> import xml.dom >>> __package__ = "xml.dom" >>> from . import minidom # Works fine >>> from ...sys import path Traceback (most recent call last): File "", line 1, in ValueError: attempted relative import beyond top-level package Either the documentation or the implementation should be fixed. |
|
|
| msg297796 - (view) |
Author: Alyssa Coghlan (ncoghlan) *  |
Date: 2017-07-06 03:58 |
| That part of the PEP was never implemented - relative imports have never been allowed to cross package boundaries in practice (which is also why "from . import sys" doesn't work as an equivalent to "import sys" at the interactive prompt or in a top level module). Rather than amending the PEP, my inclination is to drop that cross-reference from the documentation, and instead add a new subsection to https://docs.python.org/3/reference/import.html covering "Package relative imports" The key section that needs to be extracted is https://www.python.org/dev/peps/pep-0328/#guido-s-decision (minus the part about cross-package relative imports being permitted). The following section about __name__ is also worth including but needs to be updated to account for __package__ (added by PEP 366), and __spec__ (added by PEP 451). |
|
|
| msg340288 - (view) |
Author: Joannah Nanjekye (nanjekyejoannah) * |
Date: 2019-04-15 16:00 |
| I actually opened a PR with the relevant documentation suggested by @ncoghlan but what am wondering is. when @ncoghlan says That part of the PEP was never implemented - relative imports have never been allowed to cross package boundaries in practice, does it mean we dont need this to work anymore as was earlier put in the PEP? In which case this issue will be closed if my PR is merged or we should actually implement it to complete everything the PEP specified? |
|
|
| msg340299 - (view) |
Author: Brett Cannon (brett.cannon) * |
Date: 2019-04-15 18:27 |
| RE: "does it mean we dont need this to work anymore as was earlier put in the PEP?" Correct, we aren't going to implement that part of the PEP ever. |
|
|
| msg340303 - (view) |
Author: Joannah Nanjekye (nanjekyejoannah) * |
Date: 2019-04-15 19:14 |
| Great, thanks for clarity. |
|
|
| msg340780 - (view) |
Author: Alyssa Coghlan (ncoghlan) * |
Date: 2019-04-24 15:14 |
| New changeset 70bf713617e15fad390ed953e48b3c65d9bc90ec by Nick Coghlan (Joannah Nanjekye) in branch 'master': bpo-30840: Document relative imports (#12831) https://github.com/python/cpython/commit/70bf713617e15fad390ed953e48b3c65d9bc90ec |
|
|
| msg340787 - (view) |
Author: Alyssa Coghlan (ncoghlan) * |
Date: 2019-04-24 15:44 |
| New changeset 4d0233ec656bc7e7814e5f6f484e79a50a0daf91 by Nick Coghlan (Miss Islington (bot)) in branch '3.7': bpo-30840: Document relative imports (GH-12831) (GH-12938) https://github.com/python/cpython/commit/4d0233ec656bc7e7814e5f6f484e79a50a0daf91 |
|
|