[
https://issues.apache.org/jira/browse/ARROW-16070?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17514703#comment-17514703
]
Jonathan Keane commented on ARROW-16070:
----------------------------------------
It's possible they are the same, yeah. Now that I've looked it does look like
our C++ test does use brew for installing these dependencies as well (thought
it's a slightly different way of doing it that I missed the first time I
scanned the output!)
> [C++] [R] brew + aws sdk version segault
> ----------------------------------------
>
> Key: ARROW-16070
> URL: https://issues.apache.org/jira/browse/ARROW-16070
> Project: Apache Arrow
> Issue Type: Bug
> Components: C++, R
> Reporter: Jonathan Keane
> Priority: Major
>
> homebrew-r-brew nightly test failing recently with a segfault in s3 tests:
> https://github.com/ursacomputing/crossbow/runs/5745568244?check_suite_focus=true#step:8:18521
> {code}
> Start test: write_parquet with filesystem arg
> *** caught segfault ***
> address 0x0, cause 'memory not mapped'
> Traceback:
> 1: fs___FileSystem__CreateDir(self, clean_path_rel(path), isTRUE(recursive))
> 2: fs$CreateDir(minio_path("hive_dir", "group=1", "other=xxx"))
> 3: eval(code, test_env)
> 4: eval(code, test_env)
> 5: withCallingHandlers({ eval(code, test_env) if (!handled &&
> !is.null(test)) { skip_empty() }}, expectation =
> handle_expectation, skip = handle_skip, warning = handle_warning, message
> = handle_message, error = handle_error)
> 6: doTryCatch(return(expr), name, parentenv, handler)
> 7: tryCatchOne(expr, names, parentenv, handlers[[1L]])
> 8: tryCatchList(expr, names[-nh], parentenv, handlers[-nh])
> 9: doTryCatch(return(expr), name, parentenv, handler)
> 10: tryCatchOne(tryCatchList(expr, names[-nh], parentenv, handlers[-nh]),
> names[nh], parentenv, handlers[[nh]])
> 11: tryCatchList(expr, classes, parentenv, handlers)
> 12: tryCatch(withCallingHandlers({ eval(code, test_env) if (!handled &&
> !is.null(test)) { skip_empty() }}, expectation =
> handle_expectation, skip = handle_skip, warning = handle_warning, message
> = handle_message, error = handle_error), error = handle_fatal, skip =
> function(e) { })
> 13: test_code(desc, code, env = parent.frame(), reporter = reporter)
> 14: testthat::test_that(what, { skip_if(getOption("..skip.tests", TRUE),
> "arrow C++ library not available") code})
> 15: test_that("write_parquet with filesystem arg", {
> skip_if_not_available("parquet") fs$CreateDir(minio_path("hive_dir",
> "group=1", "other=xxx")) fs$CreateDir(minio_path("hive_dir", "group=2",
> "other=yyy")) expect_length(fs$ls(minio_path("hive_dir")), 2)
> write_parquet(df1, fs$path(minio_path("hive_dir", "group=1",
> "other=xxx", "file1.parquet"))) write_parquet(df2,
> fs$path(minio_path("hive_dir", "group=2", "other=yyy",
> "file2.parquet")))
> expect_identical(read_parquet(fs$path(minio_path("hive_dir",
> "group=1", "other=xxx", "file1.parquet"))), df1)})
> 16: eval(code, test_env)
> 17: eval(code, test_env)
> {code}
> Suspiciously the version of
> https://github.com/Homebrew/homebrew-core/blob/HEAD/Formula/aws-sdk-cpp.rb
> was bumped around the same time (and is a version quite a bit ahead of our
> pinned old version, though in practice this is usually fine).
--
This message was sent by Atlassian Jira
(v8.20.1#820001)