Trying to install libarchive-3.1.2-r1 with features=test fails because the tests require lrzip version 0.6 or greater. Currently lrzip >= 0.6 are all ~amd64, but libarchive is amd64 (i.e. not keyword masked). This dependency should be noted in the ebuild and the keywords should be adjusted one way or the other. Reproducible: Always Steps to Reproduce: FEATURES=test emerge libarchive Actual Results: Totals: Tests run: 42 Tests failed: 1 Assertions checked: 946 Assertions failed: 6 Skips reported: 3 Failing tests: 6: test_extract_cpio_lrz (6 failures) Details for failing tests: /var/tmp/portage/app-arch/libarchive-3.1.2-r1/temp/bsdcpio_test.2013-04-12T16.06.34-000 FAIL: bsdcpio_test ==================================================== 3 of 3 tests failed Expected Results: successful install Contents of less /var/tmp/portage/app-arch/libarchive-3.1.2-r1/temp/bsdcpio_test.2013-04-12T16.06.34-000/test_extract_cpio_lrz/test.err Attempting to work with file produced by newer lrzip version 0.6 file. Unexpected initial tag 2 in streams Fatal error - exiting bsdcpio: Child process exited with status 1: Unknown error -1 Results of eix lrzip [I] app-arch/lrzip Available versions: 0.44 ~0.611 ~0.614 Installed versions: 0.44(17:53:16 31/07/12) Homepage: http://ck.kolivas.org/apps/lrzip/README Description: Long Range ZIP or Lzma RZIP optimized for compressing large files
Should I raise this with upstream e.g. ask that they use more backwards-compatible test data?
I installed app-arch/lrzip-0.614 and tried installing libarchive again. It still has errors but they are more like this ~ # cat /var/tmp/portage/app-arch/libarchive-3.1.2-r1/temp/bsdtar_test.2013-04-12T16.18.01-000/test_extract_tar_lrz.log test_extract_tar_lrz tar/test/test_extract_tar_lrz.c:44: File should be empty: test.err File size: 34 Contents: 0000 57 61 72 6e 69 6e 67 2c 20 75 6e 61 62 6c 65 20 Warning, unable 0010 74 6f 20 73 65 74 20 6e 69 63 65 20 76 61 6c 75 to set nice valu 0020 65 0a e. I think these count as spurious errors which can be safely ignored. Presumably the inability to set nice value is a result of sandboxing and PORTAGE_NICENESS - is there any way to suppress the warning?
I think this was resolved. Feel free to reopen if you can still reproduce with 3.5.1.
