As-per title, xz-utils compiles and installs cleanly - but on attempting to build eix-0.19.1 (which requires xz-utils) I'm told that the downloaded archive is corrupt by src_unpack, despite the check-sums matching. Re-building xz-utils with FEATURES="test" results in: make[2]: Leaving directory `/usr/opt/gentoo/var/tmp/portage/app-arch/xz-utils-4.999.9_beta/work/xz-4.999.9beta/tests' make check-TESTS make[2]: Entering directory `/usr/opt/gentoo/var/tmp/portage/app-arch/xz-utils-4.999.9_beta/work/xz-4.999.9beta/tests' PASS: test_check PASS: test_stream_flags test_filter_flags.c:103: !decode(2) /opt/gentoo/bin/bash: line 5: 955012 Abort (core dumped) ${dir}$tst FAIL: test_filter_flags test_block_header.c:92: lzma_block_header_decode(&decoded_options, NULL, buf) == LZMA_OK /opt/gentoo/bin/bash: line 5: 954751 Abort (core dumped) ${dir}$tst FAIL: test_block_header PASS: test_index Good file failed: ./files/good-1-3delta-lzma2.xz FAIL: test_files.sh test_compress.sh: generated_abc../src/xz/xz: compress_generated_abc: Internal error (bug) Compressing failed: -1 compress_generated_abc FAIL: test_compress.sh ========================================= 4 of 7 tests failed Please report to lasse.collin@tukaani.org ========================================= make[2]: *** [check-TESTS] Error 1 make[2]: Leaving directory `/usr/opt/gentoo/var/tmp/portage/app-arch/xz-utils-4.999.9_beta/work/xz-4.999.9beta/tests' make[1]: *** [check-am] Error 2 make[1]: Leaving directory `/usr/opt/gentoo/var/tmp/portage/app-arch/xz-utils-4.999.9_beta/work/xz-4.999.9beta/tests' make: *** [check-recursive] Error 1 * ERROR: app-arch/xz-utils-4.999.9_beta failed: * Make check failed. See above for details. * * Call stack: * ebuild.sh, line 61: Called call-ebuildshell 'src_test' * environment, line 390: Called src_test * environment, line 2733: Called _eapi0_src_test * ebuild.sh, line 673: Called die * The specific snippet of code: * hasq test $FEATURES && die "Make check failed. See above for details."
(I was hoping to CC upstream on this Bug, but it seems that Bugzilla won't allow it. Reported separately)
anything heard from upstream?
Still working through the problems, but it looks as if the test-failures are (as with app-shells/bash) caused by some optimisation the compiler performs at its default level, -O2 (== -O). It looks as if building at -O1 or -O0 works correctly - but there's a few other (relatively minor) platform issues (misdetection of several elements in configure because of GNU/gcc assumptions and mis-reading the amount of memory available) that will hopefully be fixed shortly!
Latest code works perfectly on IRIX (inc. CPU detection and memory-size detection) so long as built at -O1 or below (e.g. not -O2/-O3 or -Ofast)
We are sorry to close this bug. We lack the man-power and devotion to support mips-irix in the tree.