retitle 1086177 liblibreoffice-java: not installable on trixie ppc64el, armel,
armhf, s390x - missing ure-java
tag 1086177 + wontfix
thanks
Hi,
Am 28.10.24 um 13:29 schrieb Lionel Elie Mamane:
regression from bookworm
ure-java is present in bookworm but not in trixie, making all of
liblibreoffice-java
libreoffice-java-common
libreoffice-nlpsolver
libreoffice-report-builder
libreoffice-wiki-publisher
libreoffice-script-provider-js
libreoffice-script-provider-bsh
uninstallable in trixie
On armel, armhf, ppc64el, s390x, yes.
Why? Because the Java bridges' code is broken (new tests found it) on armhf,
ppc64el, s390x thus in this specific case for ppc64el and s390x (armhf came
later and armel was also disabled for consistency with armhf):
libreoffice (4:7.6.3~rc2-1) experimental; urgency=medium
* New upstream release candidate
- fixes CVE-2023-6185: "Improper input validation enabling arbitrary
Gstreamer
pipeline injection"
* debian/control*in:
- add missing Build-Profiles: <!nopython>
- add missing Build-Profiles: <!nojava>
* debian/rules:
- disable Java on ppc64el, s390x and armhf. Java bridge broken,
works without Java
- fix/improve no-Java builds:
+ don't try to install dbreport config modules into
-uiconfig-report-builder if the report builder is disabled
+ switch OOO_BASE_ARCHS/OOO_REPORTBUILDER_ARCHS variables
(build base on all architectures instead of Java architectures
and report-builder-* only on archs where Java is enabled, not on
"all base archs"); there's -sdbc-{mysql,postgresql,firebird} anyway
+ don't try to build Jar_{OOoRunner,test,ConnectivityTools} if Java is
disabled as an extra safety net (instead of checking for junit only)
- remove riscv64 from OOO_CHECK_ARCHS to run only the "smoketest tests"
- bump libcmis build-dep to >= 0.6.1
- build against libcurl4-openssl-dev instead of gnutls
-- Rene Engelhard <r...@debian.org> Fri, 17 Nov 2023 07:05:17 +0100
to get a sane version into testing, with the caveat of disabling Java support
on those archs.
see also https://lists.debian.org/debian-powerpc/2023/12/msg00002.html
If we had porters actually caring and helping in the various calls for help I
did I am happy to apply a
(ideally upstreamed) patch and re-enable Java there.
(see https://lists.debian.org/debian-powerpc/2023/06/msg00053.html ff.,
continued into July and
https://lists.debian.org/debian-powerpc/2023/10/msg00026.html ff.)
Testing allowed this after hinting. And no. if you would suggest that I am not
going to make those arch-dep.
The other "solution" would be to --without-java anywhere which would be a
bigger pandoras box than open already now.
(as all that would be gone everywhere, no SRB at all, no Java extensions, ...)
-> wontfix, since there is no cantfix. Unless some porter actually does porting.
Regards,
Rene
