FTBFS because of python3.5

Bug #1506536 reported by Łukasz Zemczak on 2015-10-15
This bug affects 1 person
Affects Status Importance Assigned to Milestone
pylint (Ubuntu)
Łukasz Zemczak

Bug Description

Don't worry about the summary, it's not a python3.5 regression but more like an issue with the package handling 3.5. Happened during the wily test-rebuild [1]. Log:

/bin/sh: 3: python3.5: not found
debian/rules:15: recipe for target 'override_dh_auto_build' failed
make[1]: *** [override_dh_auto_build] Error 127
make[1]: Leaving directory '/«PKGBUILDDIR»'
debian/rules:12: recipe for target 'build' failed
make: *** [build] Error 2

[1] http://people.ubuntuwire.org/~wgrant/rebuild-ftbfs-test/test-rebuild-20151001-wily.html

Łukasz Zemczak (sil2100) wrote :

Ok, so I have a proposed fix for this failure here, but I would appreciate someone to review as I don't know enough about the conventions in python packages. The issue here is that in debian/rules we use pyversions/py3versions -s - which basically means to print all 'supported' versions, regardless if those are installed currently or not. Since right now both 3.4 and 3.5 are supported, it wants to install it for both - even though only one is installed. My proposed fix is to use -i instead - printing all supported and installed versions. This way we define what to build the package against in the dependencies. I think this is more correct.

We could also change the deps, but that would be ugly.

Daniel Holbach (dholbach) wrote :

This was fixed in xenial: https://launchpad.net/ubuntu/+source/pylint/1.4.4-1ubuntu1

Do we still want to fix this in wily?

Changed in pylint (Ubuntu):
status: In Progress → Incomplete
Steve Langasek (vorlon) wrote :

It's SRU-worthy, but note that the patch proposed in the bug is incorrect - you should never use py{,3}versions -i in a package build, packages should always be built for either the --supported set or against the --default version.

Changed in pylint (Ubuntu):
importance: Undecided → High
Sebastien Bacher (seb128) wrote :

unsubscribing sponsors, that needs to be fixed according to Steve's comment

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers