t\spec\S32-io\io-spec-win.t fails on Windows 10 with MSVS 2013

[p6rt]

Migrated from rt.perl.org#126876 (status was 'resolved')

Searchable as RT126876$

[p6rt]

From @nanis

Hello​:

Just built using http://rakudo.org/downloads/star/rakudo-star-2015.11.tar.gz.

During `nmake rakudo-spectest`, I noticed a number of test failures.
One set of these occurs in t\spec\S32-io\io-spec-win.t. Here is the
test output​:

ok 206 - updir is ".."
Method 'devnull' not found for invocant of class 'IO​::Spec'
  in block <unit> at t\spec\S32-io\io-spec-win.t​:299

# Looks like you planned 209 tests, but ran 206
Dubious, test returned 255 (wstat 65280, 0xff00)
Failed 3/209 subtests

Test Summary Report

---

t\spec\S32-io\io-spec-win.t (Wstat​: 65280 Tests​: 206 Failed​: 0)
  Non-zero exit status​: 255
  Parse errors​: Bad plan. You planned 209 tests but ran 206.
Files=1, Tests=206, 5 wallclock secs ( 0.17 usr + 0.06 sys = 0.24 CPU)
Result​: FAIL

And, here is the relevant section in the test file​:

if $*DISTRO.name !~~ any(<mswin32 netware symbian os2 dos>) {
  skip-rest 'Win32ish on-platform tests'
}
else {
  # double check a couple of things to see if IO​::Spec loaded correctly
  is IO​::Spec.devnull, 'nul', 'devnull is nul';
  is IO​::Spec.rootdir, '\\', 'rootdir is "\\"';
  ok {.IO.d && .IO.w}.(IO​::Spec.tmpdir), "tmpdir​: {IO​::Spec.tmpdir}
is a writable directory";
}

It seems somehow IO​::Spec is not delegating properly, but I haven't
investigated this. BTW, after installation, the behavior persists​:

C​:\> perl6 -e "say IO​::Spec​::Win32.devnull"
nul

C​:\> perl6 -e "say IO​::Spec.devnull"
Method 'devnull' not found for invocant of class 'IO​::Spec'
  in block <unit> at -e​:1

C​:\> perl6 -e "say IO​::Spec"
(Spec)

C​:\> perl6 -e "say IO​::Spec​::Win32"
(Win32)

And, for sanity​:

C​:\> perl6 -e "use nqp; say
nqp​::p6box_s(nqp​::atkey(nqp​::backendconfig(), 'osname'));"
MSWin32

I am afraid I do not know where to go from here, but FYI.

HTH,

-- Sinan

[p6rt]

From @nanis

I am afraid I do not know where to go from here, but FYI.

Pull request submitted. I hope this is the right place to do that​:

perl6/roast#87

The fix I propose is to invoke methods on $*SPEC rather than IO​::Spec.

HTH,

-- Sinan

[p6rt]

From @nanis

I am afraid I do not know where to go from here, but FYI.

Pull request submitted. I hope this is the right place to do that​:

perl6/roast#87

The fix I propose is to invoke methods on $*SPEC rather than IO​::Spec.

HTH,

-- Sinan

[p6rt]

From @labster

Pull request submitted. I hope this is the right place to do that​:

perl6/roast#87

I approved the pull request. If the tests pass on Windows, this ticket can be resolved.

[p6rt]

The RT System itself - Status changed from 'new' to 'open'

[p6rt]

From @labster

On Mon Dec 14 09​:29​:37 2015, labster wrote​:

Pull request submitted. I hope this is the right place to do that​:

perl6/roast#87

I approved the pull request. If the tests pass on Windows, this
ticket can be resolved.

Given the complete lack of activity/complaints in 4 months, I'm going to assume that the tests pass on Windows. Resolving ticket.

[p6rt]

@labster - Status changed from 'open' to 'resolved'