FS#56124 - [s-nail] build in clean chroot fails due to missing sudo permissions
Attached to Project:
Arch Linux
Opened by Erich Eckner (deepthought) - Thursday, 26 October 2017, 11:31 GMT
Last edited by Gaetan Bisson (vesath) - Thursday, 26 October 2017, 20:31 GMT
Opened by Erich Eckner (deepthought) - Thursday, 26 October 2017, 11:31 GMT
Last edited by Gaetan Bisson (vesath) - Thursday, 26 October 2017, 20:31 GMT
|
Details
Description:
The use of sudo in check() dos not work if the package is built via archbuild. It results in: ==> Starting check()... We trust you have received the usual lecture from the local System Administrator. It usually boils down to these three things: #1) Respect the privacy of others. #2) Think before you type. #3) With great power comes great responsibility. [sudo] password for builduser: Additional info: * package version(s) s-nail 14.9.5-1 (git revision fd3fd57127d7) Steps to reproduce: > asp export s-nail > cd s-nail > staging-x86_64-build |
This task depends upon
For many packages that makes running `make check` properly near impossible (off the top of my head: openssh, also s-nail).
Personally I build with a full environment inside a VM which does not suffer from these limitations; see: https://aur.archlinux.org/packages/buildstuff/
What would you suggest as a possible solution to this? I can add a check to see if we have sudo access and if not disable the check() function entirely. Would that work for you?
Aside: I quite like Earnestly's glibc-git package for the fact that it reimplements locale.gen as /etc/locales/en_US and other drop-in files and just builds the UTF-8 versions of all files. (Because in this day and age non-UTF-8 locales are extremely silly.) You could easily enable a locale that some package needs by adding a makedepends on some package which provides that drop-in file and runs locale-gen.
./cc-test.sh: line 2204: 2294 Segmentation fault (core dumped) ${MAILX} ${ARGS} -X'source ./.t.rc' -Xx > "${MBOX}" 2>&1
behave:atxplode-1: error: unexpected exit status: 139 != 0
behave:atxplode-1: error: checksum mismatch (got 4294967295 0)
./cc-test.sh: line 2655: 2988 Segmentation fault (core dumped) ${MAILX} ${ARGS} -ef ./.t.mbox
./cc-test.sh: line 2655: 3001 Segmentation fault (core dumped) ${MAILX} ${ARGS} -ef ./.t.mbox
./cc-test.sh: line 2655: 3002 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL @t@me ./.t.mbox
./cc-test.sh: line 2655: 3003 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL @t@you ./.t.mbox
./cc-test.sh: line 2655: 3004 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL '@>@Line 1' ./.t.mbox
./cc-test.sh: line 2655: 3005 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL '@>@Hello.' ./.t.mbox
./cc-test.sh: line 2655: 3006 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL '@>@Bye.' ./.t.mbox
./cc-test.sh: line 2655: 3007 Segmentation fault (core dumped) ${MAILX} ${ARGS} -efL '@>@Good bye.' ./.t.mbox
./cc-test.sh: line 2655: 3008 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fH ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3009 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL @t@me ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3010 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL @t@you ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3011 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL '@>@Line 1' ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3012 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL '@>@Hello.' ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3013 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL '@>@Bye.' ./.t.mbox >> "${MBOX}"
./cc-test.sh: line 2655: 3014 Segmentation fault (core dumped) ${MAILX} ${ARGS} -fL '@>@Good bye.' ./.t.mbox >> "${MBOX}" 2> /dev/null
behave:e_H_L_opts: error: checksum mismatch (got 3914351258 60)
So removing my sudo won't change much because the tests will fail under makechrootpkg. Quite frankly I'm tired of fixing these tests which assume a full-blown environment. So I guess I'll do what I did with openssh and disable the tests altogether when inside makechrootpkg.
I just wanted to let you know, that andreas_baumann had a look at the segfault, too: https://bbs.archlinux32.org/viewtopic.php?pid=474#p474