https://blogs.gentoo.org/ago/2020/07/04/gentoo-tinderbox/ Issue: sys-apps/dcfldd-1.9.2-r1 fails tests. Discovered on: amd64 (internal ref: ci) Info about the issue: https://wiki.gentoo.org/wiki/Project:Tinderbox/Common_Issues_Helper#CF0015
Created attachment 916387 [details] build.log build log and emerge --info
I think we can/should avoid the use of tee there.
Also, note that the pipes there mean we don't catch errors from dcfldd itself.
(In reply to Sam James from comment #2) > I think we can/should avoid the use of tee there. I can sure resolve the "tee /dev/stderr" but wth is "tee: /dev/stderr: Permission denied", why? :) (In reply to Sam James from comment #3) > Also, note that the pipes there mean we don't catch errors from dcfldd > itself. I can add "set -o pipefile" if that's what you're thinking of. Starting now…
> I can add "set -o pipefile" if that's what you're thinking of. pipefail of course, my bad.
Update: deciding against pipefail complexity now, the goal of the test is to detect miscompilation, all else seems out of scope to me.
inherit eapi9-pipestatus ... expected_sha256sum="$(sha256sum <<<TestInput | awk '{print $1}' \ | tee /dev/stderr ; pipestatus || die)" ... repeat for other subshell
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=a3f668d0f70f94d7c5154084703ac75355b17998 commit a3f668d0f70f94d7c5154084703ac75355b17998 Author: Sebastian Pipping <sping@gentoo.org> AuthorDate: 2025-01-12 17:46:30 +0000 Commit: Sebastian Pipping <sping@gentoo.org> CommitDate: 2025-01-12 17:46:30 +0000 sys-apps/dcfldd: Improve test for miscompilation Closes: https://bugs.gentoo.org/947985 Signed-off-by: Sebastian Pipping <sping@gentoo.org> sys-apps/dcfldd/dcfldd-1.9.2-r1.ebuild | 7 +++---- 1 file changed, 3 insertions(+), 4 deletions(-)
(In reply to Sam James from comment #7) > inherit eapi9-pipestatus > ... > expected_sha256sum="$(sha256sum <<<TestInput | awk '{print $1}' \ > | tee /dev/stderr ; pipestatus || die)" > ... > repeat for other subshell I saw that after posting but I also don't think it will work with the "|| die" /inside/ the command substitution. The example from eapi9-pipestatus.eclass has the die /outside/ the command substitution: status=$(pipestatus -v) || die "foo | bar failed, status ${status}" I didn't know that worked but it does. Interesting. Not worth the complexity though in my book. Please extend the test if you see need to (— I don't).
(In reply to Sebastian Pipping from comment #9) > (In reply to Sam James from comment #7) > > inherit eapi9-pipestatus > > ... > > expected_sha256sum="$(sha256sum <<<TestInput | awk '{print $1}' \ > > | tee /dev/stderr ; pipestatus || die)" > > ... > > repeat for other subshell > > I saw that after posting but I also don't think it will work with the "|| > die" /inside/ the command substitution. The example from > eapi9-pipestatus.eclass has the die /outside/ the command substitution: > src_prepare() { default x=$(false || die) } (bug 465008)
Black magic, I see!
