1. Forum
  2. Usenet
  3. LINUX.DEBIAN.BUGS.DIST

  • Bug#1078886: node-fstream-ignore: FTBFS: failing tests

    From =?UTF-8?B?SsOpcsOpbXkgTGFs?=@21:1/5 to All on Sat Apr 5 19:00:01 2025
    Le sam. 5 avr. 2025 à 18:06, Santiago Vila <sanvila@debian.org> a écrit :

    reopen 1078886
    thanks

    Hello. I'm really sorry to reopen this, but the recent change
    did not fix the reported problem.

    I'm attaching my build history, where 1.0.5-5 is the version
    in unstable and 1.0.5-4.1 was my internal version where I
    tested the proposed patch (which I assume it would still work,
    since the issue seems to be a concurrency problem)


    Weird, because I reproduce, and it got way better after the fix.
    I'll investigate further.



    I'm putting a bunch of failed build logs here:

    https://people.debian.org/~sanvila/build-logs/202504/

    (btw: only if it helps, I would be willing to join the team
    and fix this myself, that would also allow me to
    fix this in stable without bothering you. I have joined
    other teams for QA purposes in the past, and it seems to be
    a form of collaboration which seems to work well).


    Yes, you can, however since I already spent some time digging,
    I might as well continue to try to fix the test. If I fail to find out,
    I'll just disable them.

    <div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">Le sam. 5 avr. 2025 à 18:06, Santiago Vila &lt;<a href="mailto:sanvila@debian.org">sanvila@debian.org</a>&gt; a écrit :<br></
    <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">reopen 1078886<br>
    thanks<br>

    Hello. I&#39;m really sorry to reopen this, but the recent change<br>
    did not fix the reported problem.<br>

    I&#39;m attaching my build history, where 1.0.5-5 is the version<br>
    in unstable and 1.0.5-4.1 was my internal version where I<br>
    tested the proposed patch (which I assume it would still work,<br>
    since the issue seems to be a concurrency problem)</blockquote><div><br></div><div>Weird, because I reproduce, and it got way better after the fix.</div><div>I&#39;ll investigate further.</div><div><br></div><div>  </div><blockquote class="gmail_quote"
    style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
    I&#39;m putting a bunch of failed build logs here:<br>

    <a href="https://people.debian.org/~sanvila/build-logs/202504/" rel="noreferrer" target="_blank">https://people.debian.org/~sanvila/build-logs/202504/</a><br>

    (btw: only if it helps, I would be willing to join the team<br>
    and fix this myself, that would also allow me to<br>
    fix this in stable without bothering you. I have joined<br>
    other teams for QA purposes in the past, and it seems to be<br>
    a form of collaboration which seems to work well).<br></blockquote><div><br></div><div>Yes, you can, however since I already spent some time digging,</div><div>I might as well continue to try to fix the test. If I fail to find out, I&#39;ll just disable
    them.</div><div><br></div></div></div>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Santiago Vila@21:1/5 to All on Sat Apr 5 23:50:01 2025
    El 5/4/25 a las 23:00, Jérémy Lal escribió:
    The problem with the dh_auto_test_override is that it effectively stops pkg-js-autopkgtest from running at all.

    Hmm, why is that the case?

    To check the concurrency hypothesis, could you try version 1.0.5-4 with -j1 added after the command "tap" like this:
    tap -j1 -R tap test/00-setup.js test/common.js test/zz-cleanup.js --no-cov
    in
    debian/tests/pkg-js/test
    ?

    If I add -j1 to version 1.0.5-4 then it does not fail anymore.

    Is this the better fix than --max-parallel=1 you were looking for?

    [ btw: Yes, I think I will click on the "join team" button.
    I'd like to save you the bureaucracy of backporting the fix for
    bookworm-proposed-updates once you decide about the fix ].

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From Santiago Vila@21:1/5 to All on Sun Apr 6 00:10:02 2025
    Good news: Everything works ok in Salsa CI when adding -j1.

    Tried in private fork:

    https://salsa.debian.org/sanvila/node-fstream-ignore/-/pipelines

    Thanks.

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)
  • From =?UTF-8?B?SsOpcsOpbXkgTGFs?=@21:1/5 to All on Sun Apr 6 00:10:02 2025
    Le sam. 5 avr. 2025 à 23:40, Santiago Vila <sanvila@debian.org> a écrit :

    El 5/4/25 a las 23:00, Jérémy Lal escribió:
    The problem with the dh_auto_test_override is that it effectively stops pkg-js-autopkgtest from running at all.

    Hmm, why is that the case?


    I was surprised but it's to be expected - simply because it lacks the "--buildsystem=nodejs" argument.


    To check the concurrency hypothesis, could you try version 1.0.5-4 with
    -j1 added after the command "tap" like this:
    tap -j1 -R tap test/00-setup.js test/common.js test/zz-cleanup.js
    --no-cov
    in
    debian/tests/pkg-js/test
    ?

    If I add -j1 to version 1.0.5-4 then it does not fail anymore.

    Is this the better fix than --max-parallel=1 you were looking for?


    Yes, because it fixes the tests where they fail, at the "tap" invocation.
    I still don't understand why it doesn't fail on my sbuild setup. Maybe
    because of my /tmp/sharedmem/ partition.
    I'm pushing a new version with that fix, anyway.



    [ btw: Yes, I think I will click on the "join team" button.
    I'd like to save you the bureaucracy of backporting the fix for
    bookworm-proposed-updates once you decide about the fix ].


    Thank you.

    <div dir="ltr"><div dir="ltr"><br></div><br><div class="gmail_quote gmail_quote_container"><div dir="ltr" class="gmail_attr">Le sam. 5 avr. 2025 à 23:40, Santiago Vila &lt;<a href="mailto:sanvila@debian.org">sanvila@debian.org</a>&gt; a écrit :<br></
    <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">El 5/4/25 a las 23:00, Jérémy Lal escribió:<br>
    &gt; The problem with the dh_auto_test_override is that it effectively stops<br>
    &gt; pkg-js-autopkgtest from running at all.<br>

    Hmm, why is that the case?<br></blockquote><div><br></div><div>I was surprised but it&#39;s to be expected - simply because it lacks the &quot;--buildsystem=nodejs&quot; argument.</div><div> </div><div><br></div><blockquote class="gmail_quote" style="
    margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
    &gt; To check the concurrency hypothesis, could you try version 1.0.5-4 with -j1 added after the command &quot;tap&quot; like this:<br>
    &gt; tap -j1 -R tap test/00-setup.js test/common.js test/zz-cleanup.js --no-cov<br>
    &gt; in<br>
    &gt; debian/tests/pkg-js/test<br>
    &gt; ?<br>

    If I add -j1 to version 1.0.5-4 then it does not fail anymore.<br>

    Is this the better fix than --max-parallel=1 you were looking for?<br></blockquote><div><br></div><div>Yes, because it fixes the tests where they fail, at the &quot;tap&quot; invocation.</div><div>I still don&#39;t understand why it doesn&#39;t fail on
    my sbuild setup. Maybe because of my /tmp/sharedmem/ partition.</div><div>I&#39;m pushing a new version with that fix, anyway.</div><div><br></div><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,
    204,204);padding-left:1ex">
    [ btw: Yes, I think I will click on the &quot;join team&quot; button.<br>
       I&#39;d like to save you the bureaucracy of backporting the fix for<br>
       bookworm-proposed-updates once you decide about the fix ].<br></blockquote><div><br></div><div>Thank you.</div><div><br></div></div></div>

    --- SoupGate-Win32 v1.05
    * Origin: fsxNet Usenet Gateway (21:1/5)