Bug 1937765

Summary:	rpmbuild specifically excludes /usr/local/ from the python path
Product:	Red Hat Enterprise Linux 8	Reporter:	Pat Riehecky <riehecky>
Component:	rpm	Assignee:	Packaging Maintenance Team <packaging-team-maint>
Status:	CLOSED DUPLICATE	QA Contact:	swm-qe
Severity:	unspecified	Docs Contact:
Priority:	unspecified
Version:	CentOS Stream	CC:	bstinson, carl, jwboyer, mhroncok, pmatilai
Target Milestone:	rc	Keywords:	Triaged
Target Release:	---
Hardware:	All
OS:	Linux
Whiteboard:
Fixed In Version:		Doc Type:	If docs needed, set a value
Doc Text:		Story Points:	---
Clone Of:		Environment:
Last Closed:	2021-03-16 08:54:49 UTC	Type:	Bug
Regression:	---	Mount Type:	---
Documentation:	---	CRM:
Verified Versions:		Category:	---
oVirt Team:	---	RHEL 7.3 requirements from Atomic Host:
Cloudforms Team:	---	Target Upstream Version:
Embargoed:

Description Pat Riehecky 2021-03-11 14:19:23 UTC

Description of problem:

When building RPMs for interdependent python modules, the exclusion of /usr/local/ from the PYTHONPATH is problematic.  

In my instance, my module also requires newer versions of some modules that are packaged already within EL8.  To avoid polluting my packages, I've installed my dependencies in /usr/local/ on my build host.

When I run the package self tests, it fails due to missing modules and incorrect dependency versions.

There doesn't appear to be a macro or a switch to disable the exclude.

Version-Release number of selected component (if applicable):rpm-4.14.3-4.el8.x86_64


How reproducible: 100%


Steps to Reproduce:
1. Install some package self-test dependencies into /usr/local/ via pip
2. rpmbuild the package
3.

Actual results:

The /usr/local/ python directories are not available to python within rpmbuild

Expected results:

A way to tell rpmbuild "I do actually want to use those in this build, please leave /usr/local/ in sys.path"

If this switch exists, I'm confident the doc to describe it is not in an obvious place....

Additional info:

I believe the existing behavior is a good default that should remain the default.

Comment 1 Panu Matilainen 2021-03-11 14:48:14 UTC

Mm, which part of the build process you're talking about, specifically? rpmbuild itself doesn't know anything about python or PYTHONPATH, these would be external helpers scattered over a few different packages, and I haven't kept up with all the details. Do you mean dependencies are not generated for things in /usr/local, or something else?

Comment 2 Pat Riehecky 2021-03-11 15:17:36 UTC

In my specific case, my packages runtime deps are all satisfied by RPMs in /usr/.  However, I've got a number of deps for the self tests (%check) that are not packaged in RPMs.  I don't need the selftest support in systems using the RPM package.  

I was chatting with Miro Hrončok who mentioned

> Note that /usr/local/... is not in sys.path when RPM_BUILD_ROOT is set.

In an ideal world, I'd package up all dozen or so of these and do everything through the package set, but alas the world is not ideal...

Comment 3 Pat Riehecky 2021-03-11 21:01:37 UTC

With https://fedoraproject.org/wiki/Changes/Making_sudo_pip_safe uncovered by a few helpful folks (thanks!)

I'd say I'm probably too far off in the weeds here for RPM as I'm really trying to live a bit in and a bit outside the RPMDB.

I'm comfortable closing this out as NOTABUG or WONTFIX.

Comment 4 Miro Hrončok 2021-03-16 09:00:48 UTC


*** This bug has been marked as a duplicate of bug 1937494 ***