Closed rbuj closed 4 years ago
The build fails in fedora
Invalid timezones in ../../../data/Locations.xml.in: America/Nuuk
@raveit65
Fedora:latest has fedora 31, maybe the build will be ok in fedora 32 ?
https://hub.docker.com/_/fedora/
I don't know when they change latest, 31
to latest, 32
. Fedora 32 is released.
And i don't want to change latest
to something else only for fixing one build.
Travis Ci should be helper and not a job machine ;)
Simply ignore it.
Any way, this shows that we shouldn't use this PR for stable 1.24 branch for the moment, because fedora 31 and others use mate-1.24 too. I am sure other distros with 1.24 can also use the old location. Debian-testing is an unreleased version of debian, so they can use a patch to fix it. I don't see no reason to break fedora 31 and other released distros with stable 1.24.
Stable libmateweather-1.24.0 builds fine in fedora 32, so changing travis config is useless.
All fedora versions are using tzdata-2020a-1, see https://koji.fedoraproject.org/koji/packageinfo?packageID=50
But debian-testing use also tzdata-2020a-1
https://packages.debian.org/search?suite=bullseye&arch=any&searchon=names&keywords=tzdata
So, i am still wondering which database is used by the check of check-timezones.sh
xmllint --valid --noout ../../../data/Locations.xml.in
../../../data/check-timezones.sh ../../../data/Locations.xml.in
Invalid timezones in ../../../data/Locations.xml.in: America/Nuuk
When we push this to master than we lost option to release tarballs and update gh-pages with fedora.
My idea is to install the updated package in travis buildroot to fix fedora build. But we really need to know was changed in debian/ubuntu/archlinux ?
Anyone has an idea?
Found the problem. tzdata-2020a-1 was hanging around in updates-testing for fedora 31. https://bodhi.fedoraproject.org/updates/FEDORA-2020-f2498da5c3 Should be in stable in a few days.
All travis builds are successful.
I think most distros are using tzdata-2020a-1, including Slackware.
So , it might be save to use this with 1.24.
By the way, the build issue with older tzdata occurs only during make distcheck
and not with normal make
command.
So, i am still wondering why @sunweaver had problems to build his package for debian.
Closes #76