parallel-hashmap: update to 2.0.0

[trustbyte]

Description

Type(s)

• bugfix
• enhancement
• security fix

Tested on

macOS 26.0.1 25A362 arm64
Command Line Tools 26.0.0.0.1.1757719676

Verification

Have you

• followed our Commit Message Guidelines?
• squashed and minimized your commits?
• checked that there aren't other open pull requests for the same change?
• referenced existing tickets on Trac with full URL in commit message?
• checked your Portfile with port lint?
• tried existing tests with sudo port test?
• tried a full install with sudo port -vst install?
• tested basic functionality of all binary files?
• checked that the Portfile's most important variants haven't been broken?

[macportsbot]

Notifying maintainers:
@barracuda156 for port parallel-hashmap.

[trustbyte]

allout.txt
@barracuda156 It is working on mine (see attached) failing with the runners, any ideas? Thanks.

[barracuda156]

@trustbyte This is all I needed locally for the build to work: macos-powerpc/powerpc-ports@14c2b54
(Not sure if tests were ran or not, can check that.)

Since the project itself only asks for C++11 https://github.com/greg7mdp/parallel-hashmap/blob/88123934b46b77c3b6d80167382734cbff6eff74/CMakeLists.txt#L27 I think C++14 was due to gtest involvement. Why require C++17 though?

Upd. Perhaps gtest changed since then, and needs C++17 now.

[barracuda156]

From the CI log here:

  /opt/local/var/macports/build/parallel-hashmap-bfd57aad/work/parallel-hashmap-2.0.0/parallel_hashmap/phmap_fwd_decl.h:81:57: error: a space is required between consecutive right angle brackets (use '> >')
     81 |                             phmap::priv::Pair<const K, V>>> // alias for std::allocator
        |                                                         ^~
        |                                                         > > 
  /opt/local/var/macports/build/parallel-hashmap-bfd57aad/work/parallel-hashmap-2.0.0/parallel_hashmap/phmap_fwd_decl.h:87:53: error: a space is required between consecutive right angle brackets (use '> >')
     87 |               class Alloc = phmap::priv::Allocator<T>> // alias for std::allocator
        |                                                     ^~
        |                                                     > >
  /opt/local/var/macports/build/parallel-hashmap-bfd57aad/work/parallel-hashmap-2.0.0/parallel_hashmap/phmap_fwd_decl.h:94:63: error: a space is required between consecutive right angle brackets (use '> >')
     94 |                             phmap::priv::Pair<const Key, Value>>> // alias for std::allocator
        |                                                               ^~
        |                                                               > > 
  /opt/local/var/macports/build/parallel-hashmap-bfd57aad/work/parallel-hashmap-2.0.0/parallel_hashmap/phmap_fwd_decl.h:109:57: error: a space is required between consecutive right angle brackets (use '> >')
    109 |                             phmap::priv::Pair<const K, V>>, // alias for std::allocator
        |                                                         ^~
        |                                                         > >
  /opt/local/var/macports/build/parallel-hashmap-bfd57aad/work/parallel-hashmap-2.0.0/parallel_hashmap/phmap_fwd_decl.h:126:63: error: a space is required between consecutive right angle brackets (use '> >')
    126 |                             phmap::priv::Pair<const Key, Value>>, // alias for std::allocator
        |                                                               ^~
        |                                                               > >

I recall seeing similar complaints from gcc with some Qt4-based project, but did not investigate the matter. Does adding a space work?

[barracuda156]

@trustbyte On Catalina is fails now both from my older port and your from this PR. So this should be a draft, it is broken presently.

P. S. It fails on gmock though, not parallel-hashmap own code.

[barracuda156]

Please remove that slash, otherwise we can miss it later. And an added extra line above platfrom darwin (no need for two empty lines).

[trustbyte]

Removed, thanks for checking it up.

[barracuda156]

I wonder if we can use MacPorts’ mock instead (and then also drop dependency on git).

Upd. We got a pretty out-of-date gtest, does the current release build at all? Maybe worth fixing that first.

[barracuda156]

Okay, gtest 1.17.0 builds fine on Catalina, so perhaps something in parallel-hashmap messes it up.

main.log

[barracuda156]

@trustbyte I have opened an issue, see details there: google/googletest#4860

[barracuda156]

--->  Testing parallel-hashmap
Warning: parallel-hashmap is configured to build for the architecture(s) 'x86_64', but did not install any Mach-O files.
Executing:  cd "/opt/local/var/macports/build/_opt_CatalinaPorts_devel_parallel-hashmap/parallel-hashmap/work/build" && /usr/bin/make test 
Running tests...
/opt/local/bin/ctest --force-new-ctest-process 
Test project /opt/local/var/macports/build/_opt_CatalinaPorts_devel_parallel-hashmap/parallel-hashmap/work/build
      Start  1: test_container_memory
 1/17 Test  #1: test_container_memory ...............   Passed    0.24 sec
      Start  2: test_hash_policy_testing
 2/17 Test  #2: test_hash_policy_testing ............   Passed    0.61 sec
      Start  3: test_node_hash_policy
 3/17 Test  #3: test_node_hash_policy ...............   Passed    0.25 sec
      Start  4: test_raw_hash_set
 4/17 Test  #4: test_raw_hash_set ...................   Passed    0.61 sec
      Start  5: test_raw_hash_set_allocator
 5/17 Test  #5: test_raw_hash_set_allocator .........   Passed    0.26 sec
      Start  6: test_flat_hash_set
 6/17 Test  #6: test_flat_hash_set ..................   Passed    0.39 sec
      Start  7: test_flat_hash_map
 7/17 Test  #7: test_flat_hash_map ..................   Passed    0.95 sec
      Start  8: test_node_hash_map
 8/17 Test  #8: test_node_hash_map ..................   Passed    0.51 sec
      Start  9: test_node_hash_set
 9/17 Test  #9: test_node_hash_set ..................   Passed    0.51 sec
      Start 10: test_parallel_flat_hash_map
10/17 Test #10: test_parallel_flat_hash_map .........   Passed    1.07 sec
      Start 11: test_parallel_flat_hash_set
11/17 Test #11: test_parallel_flat_hash_set .........   Passed    0.41 sec
      Start 12: test_parallel_node_hash_map
12/17 Test #12: test_parallel_node_hash_map .........   Passed    1.03 sec
      Start 13: test_parallel_node_hash_set
13/17 Test #13: test_parallel_node_hash_set .........   Passed    0.43 sec
      Start 14: test_parallel_flat_hash_map_mutex
14/17 Test #14: test_parallel_flat_hash_map_mutex ...   Passed    1.01 sec
      Start 15: test_dump_load
15/17 Test #15: test_dump_load ......................   Passed    0.25 sec
      Start 16: test_erase_if
16/17 Test #16: test_erase_if .......................   Passed    0.25 sec
      Start 17: test_btree
17/17 Test #17: test_btree ..........................   Passed   12.23 sec

100% tests passed, 0 tests failed out of 17

Total Test time (real) =  21.01 sec

You may try this: greg7mdp/parallel-hashmap#288

[barracuda156]

I will try on Sonoma to see if I also get space errors. On Catalina the build works with default clang now. (For the record, I updated gtest to 1.17.0.)

[trustbyte]

Lets hope this one (#29666) builds and gets in first so we have something going.

[judaew]

Thank you for the PR, I just merged PR with the gtest. Now this PR needs to be rebased with upstream to check if the new gtest version resolves the issue.

How to rebase

Update your local branch with upstream:

git checkout <your_branch>
git fetch upstream
git rebase upstream/master

Resolve any merge conflicts if they occur, then force push

[trustbyte]

Just curious, why rebase? The runners use the latest master and my branch, right?
Anyway, did rebase and pushed, we'll see how it goes.

[trustbyte]

@barracuda156 It looks good, what about the compiler, should I take it out, leave at 2014 or?

Then I'll squash and done.

[barracuda156]

what about the compiler, should I take it out, leave at 2014 or?

I think it is safer to match the setting in gtest port.

[trustbyte]

OK, Left it as in gtest (

macports-ports/devel/gtest/Portfile

Line 42 in 433c023

compiler.cxx_standard 2014

), removed the git part, cleaned up. Done.

Thank you!

[barracuda156]

This should be build dependency. Alternatively, building tests can be disabled by default and moved into a variant.

[trustbyte]

Right, depends_build-append then.

[barracuda156]

Last thing – sorry, missed this – is this actually needed on some macOS version? It should be set by CMake if not by compiler to begin with.

[trustbyte]

Probably not, leftover from some test.

[barracuda156]

P. S. And if it is needed, then it is worth also reporting it to upstream so that it can be fixed by next release. I only tried on 10.6 and 10.15, neither needed manual -pthread.