X-Recipient: archive-cygwin AT delorie DOT com
DKIM-Filter: OpenDKIM Filter v2.11.0 sourceware.org 0A6E23858414
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=cygwin.com;
	s=default; t=1644168986;
	bh=hmMivYE5VHS64Osf013AUwbvAku9JD0rTKrL12QcA/c=;
	h=To:Subject:Date:References:List-Id:List-Unsubscribe:List-Archive:
	 List-Post:List-Help:List-Subscribe:From:Reply-To:From;
	b=P0S1pLKJwk81nSQNmz5PqqsdRtjrZcVVjuNW2HsKYrztqCzhb9UrL56IJgOnUHmX4
	 g/nfIArcd0odaGp8ccob5392sp5bIchV4kBmK9/03+NKoUwpvgKtX4YIykd6UPzHYF
	 qrHWgupd4654Ojq9KjGzlKisFG2jMU0rbWeJ2jWY=
X-Original-To: cygwin AT cygwin DOT com
Delivered-To: cygwin AT cygwin DOT com
DMARC-Filter: OpenDMARC Filter v1.4.1 sourceware.org 465D53858D3C
X-Injected-Via-Gmane: http://gmane.org/
To: cygwin AT cygwin DOT com
Subject: Re: python-numpy (1.22.0-1) can't be imported
Date: Sun, 06 Feb 2022 12:35:36 -0500
Message-ID: <vriuh79bq4nr.fsf@mail.aol.com>
References: <0fe2e36a-2f32-0334-0ff6-d164f56e1b48 AT gmail DOT com>
 <20220122 DOT 100435 DOT 1822912027393751150 DOT trueroad AT trueroad DOT jp>
 <89f4e748-9e1c-b0de-834b-2511b988f14a AT gmail DOT com>
 <20220122 DOT 191639 DOT 929494193658372882 DOT trueroad AT trueroad DOT jp>
Mime-Version: 1.0
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/27.2 (cygwin)
Cancel-Lock: sha1:oCsrBJNToex3FMTHbGBvZ3pKyuM=
X-Spam-Status: No, score=-2.2 required=5.0 tests=BAYES_00,
 FREEMAIL_FORGED_FROMDOMAIN, FREEMAIL_FROM, HEADER_FROM_DIFFERENT_DOMAINS,
 KAM_DMARC_STATUS, KAM_GOODAOL, SPF_HELO_NONE, SPF_PASS, TXREP,
 T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.4
X-Spam-Checker-Version: SpamAssassin 3.4.4 (2020-01-24) on
 server2.sourceware.org
X-BeenThere: cygwin AT cygwin DOT com
X-Mailman-Version: 2.1.29
List-Id: General Cygwin discussions and problem reports <cygwin.cygwin.com>
List-Archive: <https://cygwin.com/pipermail/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-request AT cygwin DOT com?subject=help>
List-Subscribe: <https://cygwin.com/mailman/listinfo/cygwin>,
 <mailto:cygwin-request AT cygwin DOT com?subject=subscribe>
From: airplanemath via Cygwin <cygwin AT cygwin DOT com>
Reply-To: airplanemath <airplanemath AT aol DOT com>
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Sender: "Cygwin" <cygwin-bounces+archive-cygwin=delorie DOT com AT cygwin DOT com>

On Sat, Jan 22 2022, Masamichi Hosoda wrote:

>> It is something like that, but "-Wl,--export-all-symbols"
>> is not used on 1.21.4 and is not needed for most of the other
>> modules on 1.22.x
>> 
>> so I am looking for a less extreme action.
>> Also to understand how it can impact other python subpackages
>
> If I understand correctly, I've found the root cause of the issue.
> I've sent a pull request to numpy.
> https://github.com/numpy/numpy/pull/20874

Based on that pull request, it looks like it's related to the patches
used to avoid python C-extension modules with only one exported function
mentioned a year ago:
https://cygwin.com/pipermail/cygwin/2021-January/247211.html
https://cygwin.com/pipermail/cygwin/2021-January/247216.html
From a glance at the Python C API docs, it looks like Python C extension
modules are supposed to have only one exported function:
https://docs.python.org/3/extending/extending.html#the-module-s-method-table-and-initialization-function
https://docs.python.org/3/extending/extending.html#providing-a-c-api-for-an-extension-module

The NumPy CFFI tests seem to want to use a few C extension modules as
normal shared libraries, which requires that the functions the tests
want to use to be exported from the DLL, but this seems to be a somewhat
unusual case.

I tend to compile my own versions of Cython (without patches), which may
have been picked up when I compiled the new version of NumPy.


-- 
Problem reports:      https://cygwin.com/problems.html
FAQ:                  https://cygwin.com/faq/
Documentation:        https://cygwin.com/docs.html
Unsubscribe info:     https://cygwin.com/ml/#unsubscribe-simple