<div dir="ltr"><div><div>Option "Publish only on success all build_lists for each arch (checking by commit hash)" can be easily enabled for each repostitory (main, contrib ...) to prevent such issues. Unfortunately if some architecture is "not mature" then all builds will fail.<br><br></div>Maybe there is a chace to modify this feature or add another one, like to choose which architectures are "mandatory" for succes build for example:<br><br><br></div><div>In repository settings add a list of architectures with checkbox to choose which one are mandatory for success build:<br></div><div>i586<br></div><div>x86_64<br></div><div>armv7hl<br></div><div>aarch64<br><br></div><div>WDYT ?<br></div><br><br></div><div class="gmail_extra"><br><div class="gmail_quote">2016-05-30 14:52 GMT+02:00 Bernhard Rosenkraenzer <span dir="ltr"><<a href="mailto:bero@lindev.ch" target="_blank">bero@lindev.ch</a>></span>:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class="">On 2016-05-30 14:50, Tomasz Gajc wrote:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Then this is a bug in urpmi which allows to install packages for different arch than chroot arch.<br>
</blockquote>
<br>
<br></span>
It's a feature to some extent -- e.g. we WANT to be able to install 32-bit wine and 32-bit compat libs etc. from the i586 repositories on x86_64 and compat libs from the armv7hl repositories on aarch64.<br>
<br>
But urpmi should be smarter about installing something for the native arch if it is available (and only look in 32-bit repos if nothing by the same name is there).<br>
<br>
Either way the out-of-sync versions should be fixed too.<br>
<br>
<br>
ttyl<br>
bero<div class="HOEnZb"><div class="h5"><br>
<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
<br>
2016-05-30 14:44 GMT+02:00 Bernhard Rosenkraenzer <<a href="mailto:bero@lindev.ch" target="_blank">bero@lindev.ch</a>>:<br>
<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
Hi,<br>
I just tried to set up a new cooker chroot -- oddly, I ended up with a number of i586 packages (e.g. m4, clang) instead of their x86_64 equivalents.<br>
<br>
Looks like urpmi unconditionally prefers the one with a newer version/release combo, and somehow (mass build of anything that failed on i586?) a few packages in the i586 tree seem to have newer release numbers.<br>
<br>
I think sooner or later we need to move to "if one architecture fails unexpectedly, the build won't be published"...<br>
<br>
ttyl<br>
bero<br>
<br>
_______________________________________________<br>
OM-Cooker mailing list<br>
<a href="mailto:OM-Cooker@ml.openmandriva.org" target="_blank">OM-Cooker@ml.openmandriva.org</a><br>
<a href="http://ml.openmandriva.org/mailman/listinfo/om-cooker_ml.openmandriva.org" rel="noreferrer" target="_blank">http://ml.openmandriva.org/mailman/listinfo/om-cooker_ml.openmandriva.org</a><br>
</blockquote>
<br>
_______________________________________________<br>
OM-Cooker mailing list<br>
<a href="mailto:OM-Cooker@ml.openmandriva.org" target="_blank">OM-Cooker@ml.openmandriva.org</a><br>
<a href="http://ml.openmandriva.org/mailman/listinfo/om-cooker_ml.openmandriva.org" rel="noreferrer" target="_blank">http://ml.openmandriva.org/mailman/listinfo/om-cooker_ml.openmandriva.org</a><br>
</blockquote>
</div></div></blockquote></div><br></div>