On Tue, 25 Jul 2023 at 11:58, Piotr Łobacz <[email protected]> wrote:
>
> Yeah, I have just checked it for acpid-src packages and diffoscope worked 
> without any errors.
>
> Question, is this test being run with acl or/and xattr? Because from what I 
> remember it was not. Can you check?
> If this will confirm, then something is still wrong with the gnu format, 
> because right now all posix - meaning with acl/xattr support - packages are 
> working for me.

Hello Piotr,

may I suggest: you need to learn to reproduce the autobuilder failures
locally. Integrators are stretched very thin, they juggle a ton of
patches (other than the ones you want to get in), and the expectation
is that when they tell you something fails, it is on you to do further
investigation. Here's how you do it:

- start by finding the failed build in the list of their definitions:
https://git.yoctoproject.org/yocto-autobuilder-helper/tree/config.json
You can see that it's this one:

        "reproducible" : {
            "MACHINE" : "qemux86-64",
            "SDKMACHINE" : "x86_64",
            "step1" : {
                "shortname" : "Reproducible Selftest",
                "EXTRACMDS" : ["${SCRIPTSDIR}/checkvnc;
OEQA_DEBUGGING_SAVED_OUTPUT=${BASE_SHAREDDIR}/pub/repro-fail/
DISPLAY=:1 oe-selftest -r reproducible"],
                "ADDLAYER" : ["${BUILDDIR}/../meta-selftest"]

            }

What that means is you need to set up a plain poky build (no extra
layers, no custom configs specific to your organization), then run
'oe-selftest -r reproducible'. That is however a heavy test as it
builds everything, so we need to narrow it down to just the failing
pieces.

- the failure log shows that reproducibility is tested via
meta/lib/oeqa/selftest/cases/reproducible.py. If you open that file,
you can see OEQA_REPRODUCIBLE_TEST_TARGET narrows down the targets
used to test reproducibility, and you can also set
OEQA_REPRODUCIBLE_TEST_PACKAGE similarly.

- now you can run oe-selftest -r reproducible. Once it finishes and
confirms the failure, head to build-st/reproducibleA and
build-st/reproducibleB, to see where the difference first appears.

You can also replicate reproducibility test by manually setting up two
different build directories with identical configs.

Alex
-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#184832): 
https://lists.openembedded.org/g/openembedded-core/message/184832
Mute This Topic: https://lists.openembedded.org/mt/100347376/21656
Group Owner: [email protected]
Unsubscribe: https://lists.openembedded.org/g/openembedded-core/unsub 
[[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-

Reply via email to