search

lizmat / ParaSeq

Smarter parallel execution of Iterables
https://raku.land/zef:lizmat/ParaSeq
Artistic License 2.0
6 stars 0 forks source link

ParaSeq testing times out with Rakudo 2025.02 #4

Open japhb opened 1 month ago

japhb commented 1 month ago

Trying to "build the world" at Rakudo 2025.02, I can't install App::Rak::Complete because ParaSeq:ver<0.2.7>:auth<zef:lizmat> fails testing. It appears to be fine until passing 10-invert, and then it goes very low CPU and just stalls out for a while until eventually the harness kills it:

===> Testing: ParaSeq:ver<0.2.7>:auth<zef:lizmat>
[ParaSeq] t/01-basic.rakutest ........... ok
[ParaSeq] t/02-reverse.rakutest ......... ok
[ParaSeq] t/03-elems.rakutest ........... ok
[ParaSeq] t/04-tail.rakutest ............ ok
[ParaSeq] t/05-head.rakutest ............ ok
[ParaSeq] t/06-skip.rakutest ............ ok
[ParaSeq] t/07-first.rakutest ........... ok
[ParaSeq] t/08-grep.rakutest ............ ok
[ParaSeq] t/09-map.rakutest ............. ok
[ParaSeq] t/10-invert.rakutest .......... ok
===> Testing [FAIL]: ParaSeq:ver<0.2.7>:auth<zef:lizmat>
Aborting due to test failure: ParaSeq:ver<0.2.7>:auth<zef:lizmat> (use --force-test to override)
"zef" unexpectedly returned exit value 1 at /usr/share/perl5/IPC/System/Simple.pm line 578, <$_[...]> line 75.
    IPC::System::Simple::_check_exit("zef", 1, ARRAY(0x555f36669948)) called at /usr/share/perl5/IPC/System/Simple.pm line 550
    IPC::System::Simple::_process_child_error(256, "zef", ARRAY(0x555f36669948)) called at /usr/share/perl5/IPC/System/Simple.pm line 217
    IPC::System::Simple::runx(ARRAY(0x555f36669948), "zef", "-v", "install", "App::Rak::Complete") called at /usr/share/perl5/IPC/System/Simple.pm line 168
    IPC::System::Simple::run("zef", "-v", "install", "App::Rak::Complete") called at (eval 7) line 12
    eval {...} called at (eval 7) line 11
    Fatal::__ANON__("zef", "-v", "install", "App::Rak::Complete") called at /home/geoff/bin/install-raku-modules line 32
    main::zef_install("App::Rak::Complete") called at /home/geoff/bin/install-raku-modules line 21
    main::interpret_list_file("/home/geoff/bin/raku-module-list", "*main::zef_install") called at /home/geoff/bin/install-raku-modules line 40
    main::MAIN() called at /home/geoff/bin/install-raku-modules line 44
 at /home/geoff/bin/install-raku-modules line 32

For clarity, note that install-raku-modules is a simple bootstrapping Perl 5 script that at this stage in the world-building process is just calling zef -v install foo in a loop via multi-argument system() under the influence of use autodie ':all'.

(Why not switch to a Raku script at that point in the bootstrapping, since if zef is working, Rakudo clearly is? To reduce the chance of influencing module testing results because of an unrelated Rakudo IPC bug as of some random Rakudo commit, thus causing me to file bogus install failure issues.)

japhb commented 1 month ago

From IRC discussion:

13:19 < [Coke]> (just kicked off a zef --serial install of R::A::C and it also appears to.... slow down on ParaSeq
13:23 < japhb> [Coke], lizmat: Clarified in https://github.com/lizmat/ParaSeq/issues/4
14:03 < [Coke]> Oh, I forgot about this - it's still hanging. :)
14:04 < [Coke]> I'm on macOS 14.6.1 here
14:04 < [Coke]> killed it by hand.
14:05 < japhb> OK, thanks for the confirmation that it's not just me.  :-)
14:05 < japhb> Mind if I add this to the bug report?
14:07 < [Coke]> that's fine
14:08 < [Coke]> while running *11, the last output here is [ParaSeq] ok 3 - The object is-a 'ParaSeq'
14:09 < [Coke]> (did an install with -serial, and then a zef look on paraseq and a zef test -v .