Re: git: 329f5c880471 - main - ports-mgmt/reprise: unbreak bulk -a

From: Mark Millard <marklmi_at_yahoo.com>
Date: Wed, 13 Apr 2022 21:39:29 UTC
[A resend because I managed to not include
dev-commits-ports-main or freebsd-ports the
fist time. I also dropped a CC to someone
that would have already got the original.]

From: Dmitry Marakasov <amdmi3_at_amdmi3.ru> 
Date: Wed, 13 Apr 2022 16:14:02 +0300 :
* Antoine Brodin (antoine_at_FreeBSD.org) wrote:

> Please revert this.

Until it no longer leads to bulk -a breaking, that would
stop all build on the FreeBSD build servers affected: no
more package updates for any packages from those servers.
(I've no clue if it is all platforms or not that have
bulk -a broken before this change. The details of the
bulk -a breaking are not clear. It is not even clear if
only bulk -a would fail for poudriere use.)

In such a context, I'd expect the sequence for handling
the issue(s) to keep building of ports into packages
on the FreeBSD build servers operational until the problem
for ports-mgmt/reprise referencing 3.10 is handled. That
need not mean that fixing the problem is being avoided.
(Not that I've any view into the details of the handling.)

> You're removing _correct_ python version limitation

Well, if one is doing builds oneself and sets the python3
and python default for the environment to 3.10 , the IGNORE
would not happen if I understand the change right. (But
other ports might reject 3.10 as too new.) So:

DEFAULT_VERSIONS+= python3=3.10 python=3.10

in the likes of /etc/make.conf or, for poudriere, an
appropriate /usr/local/etc/poudriere.d/*make.conf .

> and preventing
> port users from building the port.

Only if they need the defaults to be < 3.10 for some other
reason, if I understand right.

> Also the IGNORE is incorrect for
> it will never support outdated versions of python.

I agree that the "yet" in the wording looks to be odd unless
it was known that there was an effort to support older python
versions. "yet" would be more common for rejections of newer
versions.

> If it breaks poudriere, it should be fixed in poudriere,

What was done does not prevent such, if the problem is even
in poudriere itself. But, until bulk -a is working, it seems
unlikely that FreeBSD would choose to stop updating packages
via the FreeBSD build servers until then.

> or the
> port should be marked in a way so poudriere does not build it, not
> affecting port users.

It looks to me like:

DEFAULT_VERSIONS+= python3=3.10 python=3.10

may be a way forward for now.

>> The branch main has been updated by antoine:
>> 
>> URL: https://cgit.FreeBSD.org/ports/commit/?id=329f5c8804717f22745217a686ec4c115a5a40d4
>> 
>> commit 329f5c8804717f22745217a686ec4c115a5a40d4
>> Author:     Antoine Brodin <antoine_at_FreeBSD.org>
>> AuthorDate: 2022-04-12 07:48:55 +0000
>> Commit:     Antoine Brodin <antoine_at_FreeBSD.org>
>> CommitDate: 2022-04-12 07:48:55 +0000
>> 
>>    ports-mgmt/reprise: unbreak bulk -a
>> ---
>> ports-mgmt/reprise/Makefile | 10 ++++++++--
>> 1 file changed, 8 insertions(+), 2 deletions(-)
>> 
>> diff --git a/ports-mgmt/reprise/Makefile b/ports-mgmt/reprise/Makefile
>> index 9036d653355c..b2e81455ecb0 100644
>> --- a/ports-mgmt/reprise/Makefile
>> +++ b/ports-mgmt/reprise/Makefile
>> _at__at_ -12,7 +12,7 _at__at_ RUN_DEPENDS=	${PYTHON_PKGNAMEPREFIX}jsonslicer>=0:textproc/py-jsonslicer_at_${PY_FL
>> TEST_DEPENDS=	${PYTHON_PKGNAMEPREFIX}pytest>=0:devel/py-pytest_at_${PY_FLAVOR} \
>> 		${PYTHON_PKGNAMEPREFIX}pytest-asyncio>=0:devel/py-pytest-asyncio_at_${PY_FLAVOR}
>> 
>> -USES=		python:3.10+
>> +USES=		python
>> USE_GITHUB=	yes
>> GH_ACCOUNT=	AMDmi3
>> USE_PYTHON=	autoplist distutils noflavors
>> _at__at_ -29,4 +29,10 _at__at_ post-install-DOCS-on:
>> do-test:
>> 	_at_cd ${TEST_WRKSRC} && ${SETENV} ${TEST_ENV} ${PYTHON_CMD} -m pytest -v -rs
>> 
>> -.include <bsd.port.mk>
>> +.include <bsd.port.pre.mk>
>> +
>> +.if ${PYTHON_REL} < 31000
>> +IGNORE=		does not yet support python < 3.10
>> +.endif
>> +
>> +.include <bsd.port.post.mk>
>> 
> 

===
Mark Millard
marklmi at yahoo.com


===
Mark Millard
marklmi at yahoo.com