Marius Bakke writes: > Marius Bakke writes: > >> Maxim Cournoyer writes: >> >>> Hello Guix! >>> >>> While attempting to "guix environment guix" after doing a "git pull" >>> in my guix checkout (master), it attempted to build many packages >>> (bash, gcc, etc.) and failed building `serf' due to test failures: >>> >>> ............F.FFF.FFF.FF.FFFFF................................... >>> >>> There were 14 failures: >>> 1) test_ssl_trust_rootca: test/test_util.c:438: expected <0> but was <120199> >>> 2) test_ssl_certificate_chain_with_anchor: test/test_util.c:438: >>> expected <0> but was <120199> >>> 3) test_ssl_certificate_chain_all_from_server: test/test_util.c:438: >>> expected <0> but was <120199> >>> 4) test_ssl_no_servercert_callback_allok: test/test_util.c:438: >>> expected <0> but was <120170> >>> 5) test_ssl_large_response: test/test_util.c:438: expected <0> but was <120170> >>> 6) test_ssl_large_request: test/test_util.c:438: expected <0> but was <120170> >>> 7) test_ssl_client_certificate: test/test_util.c:438: expected <0> but >>> was <120170> >>> 8) test_ssl_future_server_cert: test/test_util.c:438: expected <0> but >>> was <120199> >>> 9) test_setup_ssltunnel: test/test_util.c:438: expected <0> but was <120170> >>> 10) test_ssltunnel_basic_auth: test/test_context.c:2116: expected <0> >>> but was <120170> >>> 11) test_ssltunnel_basic_auth_server_has_keepalive_off: >>> test/test_context.c:2116: expected <0> but was <120170> >>> 12) test_ssltunnel_basic_auth_proxy_has_keepalive_off: >>> test/test_context.c:2116: expected <0> but was <120170> >>> 13) test_ssltunnel_basic_auth_proxy_close_conn_on_200resp: >>> test/test_context.c:2116: expected <0> but was <120170> >>> 14) test_ssltunnel_digest_auth: test/test_util.c:438: expected <0> but >>> was <120170> >>> >>> !!!FAILURES!!! >>> Runs: 65 Passes: 51 Fails: 14 >> >> The SSL certificates in "test/server" expired 18/04. >> >> https://svn.apache.org/viewvc?view=revision&revision=1792234 >> >> I'll try to port this patch, but "viewvc" can't give binary diffs, so >> will have to clone it locally first. > > Apparently "patch" does not yet support binary diffs either: > > patching file test/test_buckets.c > Hunk #1 succeeded at 1232 (offset -91 lines). > patching file test/server/serfcacert.pem > File test/server/serfclientcert.p12: git binary diffs are not supported. > patching file test/server/serfrootcacert.pem > patching file test/server/serfserver_expired_cert.pem > patching file test/server/serfserver_future_cert.pem > patching file test/server/serfservercert.pem > patching file test/server/serfserverkey.pem > source is under 'serf-1.3.8' > applying '/gnu/store/qbh97afjjgkn332f8ncn6k8pg74wfc8w-serf-comment-style-fix.patch'... > applying '/gnu/store/vk533j8s56nhlzz2hnqn853xb9yl6b6y-serf-update-test-certificates.patch'... > builder for `/gnu/store/3kcz9as8i986fxqj368n6cg3k7nhj7cq-serf-1.3.8.tar.xz.drv' failed to produce output path `/gnu/store/n4479kx0qx8rblzqnwgxg7jnp1x1xyjh-serf-1.3.8.tar.xz' > > I'll try to run `git apply` in a phase instead, although that will > almost double the reverse dependencies of "git". Never mind, "serf" is already a dependency of git (through "subversion"), so this did not work very well.. Perhaps we could have a "git-minimal" for cases like this? I suppose that could be useful for "git-fetch" as well. I'll give that a go shortly, unless there are better suggestions.