We currently have ~170 failing jobs on hydra:
https://hydra.nixos.org/jobset/nixos/trunk-combined
We're hoping for a big effort from all contributors and maintainers to get this down to zero for the release at the end of September.
@fpletz and I will be updating this issue with further information continuously.
Off we go on a sprint to making 17.09 the best release possible!
Getting latest evaluation for http://hydra.nixos.org/jobset/nixos/release-17.09
Parsing evaluation https://hydra.nixos.org/eval/1395311
Technical proposal: «Progress on: #28643» in PRs if someone doesn't like putting a mere issue link in motivation. «Fixes» makes GitHub close the issue, which is not desired.
fbterm
: fixed slightly before this issue was posted
allegro
: works in a local sandbox build from master
Hm, «no maintainer» condition for such a list is not search-friendly.
cuneiform
: bumped/fixed
I think linuxPackages.rtl8723bs
should be removed. According to hadess/rtl8723bs@3bb1d33ad98c1d77b5e17a1b707e137b35e6d0a5 the driver is now included in the kernel (albeit only for 4.12, maybe we should update and call it broken for != 4.9?)
Edit: I am marking the package broken for 4.12+ and fixing it for 4.9.
Cannot figure out nixpkgs.linuxPackages.pktgen.x86_64-linux
, got to this point and not sure what is wrong with the Makefile
...
[...]
== common
CC copyright_info.o
CC port_config.o
CC core_info.o
CC cmdline_parse_args.o
CC lscpu.o
CC utils.o
CC coremap.o
CC _pcap.o
CC cksum.o
CC l2p.o
CC pg_inet.o
AR libcommon.a
SYMLINK-FILE include/cksum.h
/nix/store/axldd4qjamb2c1lgi7d6k3s51qyy4610-bash-4.4-p12/bin/bash: /nix/store/4lqnmvmfaxy6bl279v5a3b05rjkzad6w-dpdk-17.05.1-4.9.45/buildtools/relpath.sh: No such file or directory
ln: '/tmp/nix-build-pktgen-3.4.0.drv-0/pktgen-3.4.0/lib/common/x86_64-native-linuxapp-gcc/include' and './include' are the same file
ChromiumOS and pktgen
are mysteries to me. All other kernel modules should be fixed with the above PRs.
xxdiff-tip can be marked as done.
Some of the mentions (including mine) don't seem to be getting recognized. Does GitHub have some sort of cap on the number of users that can be mentioned at once? I count 49 unique users that were successfully mentioned.
lincity_ng
: bumped to a working version.
asc
: bumped to a working version.
swift
: Build fails due to our clang stdenv no longer working: #28223. Should get that fixed for 17.09 as well, will look into these soon-ish.
clementineFree
is now fixed, no thanks to me.
atlasWithLapack
: I removed this because it is no longer used in Nixpkgs. We should prefer OpenBLAS anyway.
libsForQt56.prison
: I marked this package broken because it is not expected to build with Qt 5.6.
Maybe it's just me but I feel we should be more liberal with removing some of these packages. Some of them haven't built successfully in a long time and/or been updated in 2+ years...
xconq
: fixed
xmacro
: bumped, fixed
obliv-c
: bumped, fixed
afl
: made an issue https://github.com/NixOS/nixpkgs/issues/28712
notmuch-addrlookup is waiting for new release for notmuch-0.25 compatibility,
https://github.com/aperezdc/notmuch-addrlookup-c/commit/88f156d04990a71c6ad6fc2757b537b44e3c4d00
/cc @adamruzicka
hivemind/hivemindd should now be fixed: https://github.com/NixOS/nixpkgs/pull/28731
nilfs-utils
: fixed
notmuch-addrlookup
should be fixed: #27691
arm-frc-linux-gnueabi-binutils
currently being worked on. Binutils is always a bit of a pain :/
special thanks to @disassembler for fixing all the gcc6 breakages.
gnome-chess, gnome-dictionary and gnome-shell build.
I'm still getting hydra failure emails, but the build it's complaining about doesn't include the fix that was merged. Is there anything else I need to do?
@canndrew: the last Hydra builds are after the merge.
helmholtz builds successfully.
hhvm builds successfully.
irrlicht builds successfully
smugline, pokerth, pokerth-server fixed (c91b4fe989, 52dc8f227f)
no errors with lcdproc
Xiphos needs a new version of "libgtkhtml".
Working on it... Tried upgrading to 4.x which compiled but broke other
packages. I'll add a gtkhtml4 library tonight...
I also have a new xiphos build I haven't pushed yet that builds from latest
GitHub release that needs some more deps. I'll try to get it pushed
tomorrow.
New version of namecoin complains about Boost::System.
Further to @disassembler's note - I have just tried building lcdproc as well and it's fine.
Just fixed dbus-broker.
bepasty-server got fix with 63b66bf960d4e8749349bb593a776d472caaee88
Just updated the list to the latest evaluation of 17.09
I fixed the build for py3status
in #28976 but apparently the package is not really working to begin with...
FYI, herqq released 2.1.0 for me. I re-did that PR this morning and it's ready to go.
I have a fix for pywbem
but it needs a little more testing before merging.
After a few hours of bisecting it seems the failure in vapoursynth
is caused by the gcc 6 update.
Do you know anything about this error?
C++ compiler does not understand std::mutex (a C++11 feature)
Lighning runs OK against master.
@colescott any update on arm-frc-linux-gnueabi-binutils?
I will get mcomix
to fix.
Sorry for the late reply.
Fixing odp-dpdk and pktgen and disabling ofp.
Pull request for master is pending:
https://github.com/NixOS/nixpkgs/pull/29100
Once it's merged I will send a PR for 17.09 branch.
Regards,
Ruslan
On Fri, Sep 8, 2017 at 7:23 AM, Anderson Torres notifications@github.com
wrote:
I will get mcomix to fix.
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
https://github.com/NixOS/nixpkgs/issues/28643#issuecomment-328116943,
or mute the thread
https://github.com/notifications/unsubscribe-auth/ACNqaNfE8T6TK7DLUfSaN-K8ntVdkNoAks5sgU3lgaJpZM4PE6iE
.
I've fixed openbrf
and python.pkgs.xgboost
. Warning: the latter was because GFortran wasn't kept in sync with gcc
so we used GFortran 5 and GCC 6. I've defaulted gfortran = gfortran6
now; tried to build R
and python.pkgs.xgboost
with it and it works. Still, I'll keep in check what Hydra says about anything GFortran 6 breaks.
mcrl2 will be fixed by https://github.com/NixOS/nixpkgs/pull/29199
@Mic92 I fixed the build but now I am having issues with the libgcc for the linker. I'll try some more stuff this week.
lxqtsession was fixed by #29132
About jscoverage. Newer gcc is picking up an error about casting int to pointer. However the value in question is always 0, so it's not really an error. Do we (a) patch the source, (b) adjust the CFLAGS, or (c) delete the package (it has been deprecated in favour of a java implementation). cc @chaoflow
python36Packages.libnacl: I cannot reproduce the failure locally neither in master nor in staging.
Also, mentions don't work properly.
@rvl I just fixed it changing JS_FALSE -> NULL.
@disassembler OK option (a) then, good stuff :star: :+1:
freeswitch will be fixed by #29463
Can someone cherry-pick #29258 into release-17.09?
@rnhmjoj done
I tested my changes to the radicale tests in #29521 locally when I made the PR and just again now with a fresh master, and nix-build nixos/tests/radicale.nix
is succeeding for me. Would appreciate some help figuring out why it's failing on Hydra (cc @Infinisil and @pSub as co-maintainers).
I've been testing on Arch Linux with Nix 1.11.15, sandboxing enabled,
@globin I'm unsure if anyone aside of me has used the ltc_scrypt
package. It is a dependency of electrum-ltc
, which I myself don't use anymore.
So I'd be fine if it was removed.
electrum-ltc
still builds actually; perhaps the package only won't support python3 or something...
PR #29763 disable Hydra builds of purePackages.gen. Not sure if anything else needs to be done with this in terms of this issue.
The remaining packages will be marked as broken, thanks for your effort!
Most helpful comment
Maybe it's just me but I feel we should be more liberal with removing some of these packages. Some of them haven't built successfully in a long time and/or been updated in 2+ years...