ARROW-10392: [C++][Gandiva] Avoid string copy while evaluating IN expression

[pprudhvi]

No description provided.

[pprudhvi]

@github-actions crossbow submit -g gandiva

[github-actions]

https://issues.apache.org/jira/browse/ARROW-10392

[github-actions]

Revision: 84515e88c39a6ae20ffc9ae86c4eb02273a6aa70

Submitted crossbow builds: ursa-labs/crossbow @ actions-680

Task	Status
gandiva-jar-osx
gandiva-jar-xenial

[pprudhvi]

still couldn't avoid string copy - the copy is now happening inside HashTable

#0  0x00007f8b9e690159 in __lll_unlock_wake_private () from /lib64/libc.so.6
#1  0x00007f8b9e60df78 in _L_unlock_14902 () from /lib64/libc.so.6
#2  0x00007f8b9e60afe5 in malloc () from /lib64/libc.so.6
#3  0x00007f8b9d3d20cd in operator new(unsigned long) () from /lib64/libstdc++.so.6
#4  0x00007f8b9d430c79 in std::string::_Rep::_S_create(unsigned long, unsigned long, std::allocator<char> const&) () from /lib64/libstdc++.so.6
#5  0x00007f8b9d432531 in char* std::string::_S_construct<char const*>(char const*, char const*, std::allocator<char> const&, std::forward_iterator_tag) () from /lib64/libstdc++.so.6
#6  0x00007f8b9d4325ed in std::basic_string<char, std::char_traits<char>, std::allocator<char> >::basic_string(char const*, unsigned long, std::allocator<char> const&) () from /lib64/libstdc++.so.6
#7  0x00007f8b595c7c4a in std::_Hashtable<nonstd::sv_lite::basic_string_view<char, std::char_traits<char> >, nonstd::sv_lite::basic_string_view<char, std::char_traits<char> >, std::allocator<nonstd::sv_lite::basic_string_view<char, std::char_traits<char> > >, std::__detail::_Identity, std::equal_to<nonstd::sv_lite::basic_string_view<char, std::char_traits<char> > >, std::hash<nonstd::sv_lite::basic_string_view<char, std::char_traits<char> > >, std::__detail::_Mod_range_hashing, std::__detail::_Default_ranged_hash, std::__detail::_Prime_rehash_policy, std::__detail::_Hashtable_traits<false, true, true> >::count(nonstd::sv_lite::basic_string_view<char, std::char_traits<char> > const&) const () from /data/yarn/data/usercache/dremio/appcache/application_1603758047742_0001/container_1603758047742_0001_01_000003/tmp/libgandiva_jni.so87f06318-3d22-4bda-be14-ef2458081d87
#8  0x00007f8b595c7258 in gdv_fn_in_expr_lookup_utf8 () from /data/yarn/data/usercache/dremio/appcache/application_1603758047742_0001/container_1603758047742_0001_01_000003/tmp/libgandiva_jni.so87f06318-3d22-4bda-be14-ef2458081d87

@bkietz , any reason why the copy is happening even when we are looking for a string_view in hashtable<string_view>?

[pprudhvi]

Turns out string_view hash function internally creates a string
https://github.com/martinmoene/string-view-lite/blob/master/include/nonstd/string_view.hpp#L1583

[pprudhvi]

@github-actions crossbow submit -g gandiva

[github-actions]

Revision: eec6dfd

Submitted crossbow builds: ursa-labs/crossbow @ actions-682

Task	Status
gandiva-jar-osx
gandiva-jar-xenial

[bkietz]

This looks good to me, any idea what is causing the linker errors?

[pprudhvi]

MinGW link problem is caused by msys2/MINGW-packages#7170 .
It'll be fixed in upstream.

#8521 (comment)