IRC channel logs


back to list of logs

<mark_weaver>I've been building various things on core-updates using x86_64 as well, and although I successfully built new x86_64 bootstrap-tarballs (and also all of the packages I patched for loongson), I've noticed that both qemu and gnutls now FTBFS on core-updates using x86_64.
<mark_weaver>gnutls fails some tests, and qemu tries to use bison and flex, which aren't inputs.
<mark_weaver>anyway, I have to switch my YeeLoong to the newer (and buggier) kernel in order to build m4. bbiab
<civodul>howdy, mark_weaver
<mark_weaver>hi civodul!
<civodul>just reading what you wrote
<civodul>weird that QEMU fails in core-updates and not in master
<mark_weaver>my guess is that the timestamps on the pregenerated flex/bison files are somehow becoming older than the source files they are generated from.
<civodul>actually it's GLib that fails to build:
<mark_weaver>so it could be the result of a change in how patching is done.
<civodul>QEMU builds on 32-bit tho:
<civodul>no, because the patching change was done some time ago and didn't lead to any failure back then
<civodul>hmm the GLib log above doesn't show anything
<civodul>must have been a transient error *sigh*
<civodul>i've restarted the build, so we'll see
<civodul>i'm busy now but i'll try to look at it tonight
<mark_weaver>glib built successfully on my x86_64 using core-updates
<civodul>though i'll be somewhat busy as well ;-)
<mark_weaver>but qemu failed
<civodul>must be a non-deterministic failure
<civodul>test failure?
<mark_weaver>well, it's not a high priority. the bootstrap tarballs seems to be fine, at least on x86_64, and hopefully they'll build on loongson as well.
<mark_weaver>qemu fails because it tries to use flex and bison, which are not inputs.
<civodul>ah right, sorry
<mark_weaver>gnutls has test failures
<mark_weaver>this is all on x86_64. my yeeloong hasn't gotten that far yet :)
<mark_weaver>the gnutls test failures are in gnutls's guile directory (the core passes fine). the failures are: session-record-port.scm and x509-auth.scm.
<mark_weaver>I can make build logs available if you like.
<civodul>mark_weaver: that reminds me something that was discussed recently on the gnutls list
<civodul>and fixed
<civodul>will see later