Closed p5pRT closed 8 years ago
Blead perl on WinXP 32 bits with VC 2003. Git reverting commit "VAX: test changes for VAX floats" causes op/sprintf.t to pass.
------------------------------------------------------ ok 228 - >12345.7\< ok 229 - >+12345.7\< ok 230 - >12345.7\< ok 231 - >0 MISSING\< ok 232 - > 0 MISSING\< ok 233 - >0 MISSING\< ok 234 - >0C MISSING\< not ok 235 - >%.0g\< >-0.0\< >-0\< >0\< C99 standard mandates minus sign but C89 does not ok 236 - >1e+04\< >1e+004\< three-digit exponent accepted ok 237 - >1.e+04\< >1.e+004\< three-digit exponent accepted ok 238 - >1.2e+04\< >1.2e+004\< three-digit exponent accepted ok 239 - >1.2e+04\< >1.2e+004\< three-digit exponent accepted ok 240 - >12345.6789\< ok 241 - > 12345.6789\< ok 242 - >0012345.6789\< ------------------------------------------------------ ok 548 - >030:031:032\< perl \#83194: vector flag + custom separator + dynamic precision ok 549 - >118.101.114.115.105.111.110\< perl \#102586: vector flag + "version" ok 550 - > 011/ 022/ 033\< four reordered args Failed 1/550 subtests
Test Summary Report ------------------- op/sprintf.t (Wstat: 0 Tests: 550 Failed: 1) Failed test: 235 Files=1\, Tests=550\, 0 wallclock secs ( 0.11 usr + 0.00 sys = 0.11 CPU) Result: FAIL ------------------------------------------------------
This refers to commit 15899733e6c3ba2d82e9b5373dabc6958554975c\, specifically to this test:
->%.0g\< >-0.0\< >-0\< >C99 standard mandates minus sign but C89 does not skip: MSWin32 VMS hpux:10.20 openbsd netbsd:1.5 irix darwin freebsd:4.9 android\< +>%.0g\< >-0.0\< >-0\< >C99 standard mandates minus sign but C89 does not skip: MSWin32 VMS netbsd:vax-netbsd hpux:10.20 openbsd netbsd:1.5 irix darwin freebsd:4.9 android\<
But really this change:
- } elsif ($os =~ /\b$^O(?::(\S+))?\b/i) { + } elsif ($os =~ /\b$^O(?::(\S+))\b/i) {
Which makes the skip-by-os only work with a :version or :archname. The patch which I will attach momentarily restores the check for plain "OS".
There's another issue as well\, though. The problem commit tried to add support for skip-by-archname\, and now skips certain tests for 'netbsd:vax-netbsd' - but what about 'netbsd:vax-netbsd-thread-multi'? I'll attach a second patch that fixes that.
-- Respectfully\, Dan Collins
The RT System itself - Status changed from 'new' to 'open'
On Fri Jul 15 15:59:53 2016\, bulk88 wrote:
Blead perl on WinXP 32 bits with VC 2003. Git reverting commit "VAX: test changes for VAX floats" causes op/sprintf.t to pass.
This appears to have been fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d.
Tony
On Fri Jul 15 17:49:18 2016\, dcollinsn@gmail.com wrote:
There's another issue as well\, though. The problem commit tried to add support for skip-by-archname\, and now skips certain tests for 'netbsd:vax-netbsd' - but what about 'netbsd:vax-netbsd-thread-multi'? I'll attach a second patch that fixes that.
Should the comment above that\, and before the __END__ also be updated?
Tony
On Sun Jul 17 17:52:43 2016\, tonyc wrote:
On Fri Jul 15 15:59:53 2016\, bulk88 wrote:
Blead perl on WinXP 32 bits with VC 2003. Git reverting commit "VAX: test changes for VAX floats" causes op/sprintf.t to pass.
This appears to have been fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d.
Tony
It wasn't fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d
On\, SHA-1: 9648eab63b366c001d38f6de565f005cf6cea10a
* Add epigraphs for 5.22.3-RC2 and 5.24.1-RC2
I get
C:\perl521\src\win32>cd ..\t & perl harness -v op/sprintf.t & cd ..\win32 # Failed test 235 - >%.0g\< >-0.0\< >-0\< >0\< C99 standard mandates minus sign but C89 does not at op/sprintf.t line 168 op/sprintf.t .. 1..550 ok 1 - >%6. 6s INVALID REDUNDANT\< (See use of $w in code above) ok 2 - >%6 .6s INVALID REDUNDANT\< ok 3 - >%6.6 s INVALID REDUNDANT\< ok 4 - skip %A tested in sprintf2.t\, failure expected on MSWin32 5.1 MSWin32-x86-multi-thread ok 5 - >11111111111111111111111111111111\< ok 6 - >11111111111111111111111111111111\< ok 7 - >0B11111111111111111111111111111111\< ok 8 - >%C INVALID REDUNDANT\< ok 9 - >2147483647\< Synonym for %ld ok 10 - >1.234568E+05\< >1.234568E+005\< three-digit exponent accepted ok 11 - >123456.789000\< Synonym for %f ok 12 - >1.23457E+06\< >1.23457E+006\< three-digit exponent accepted ok 13 - >1.23457E+102\< ok 14 - >1.23457E-102\< ok 15 - >12345.7\< ok 16 - >1.23457E+102\< exponent too big ok 17 - >1.23457E-102\< exponent too small ok 18 - >%H INVALID REDUNDANT\< ok 19 - >%I INVALID REDUNDANT\< ok 20 - >%J INVALID REDUNDANT\< ok 21 - >%K INVALID REDUNDANT\< ok 22 - >%L INVALID REDUNDANT\< ok 23 - >%M INVALID REDUNDANT\< ok 24 - >%N INVALID REDUNDANT\< ok 25 - >37777777777\< Synonym for %lo ok 26 - >%P INVALID REDUNDANT\< ok 27 - >%Q INVALID REDUNDANT\< ****cut***** ok 229 - >+12345.7\< ok 230 - >12345.7\< ok 231 - >0 MISSING\< ok 232 - > 0 MISSING\< ok 233 - >0 MISSING\< ok 234 - >0C MISSING\< not ok 235 - >%.0g\< >-0.0\< >-0\< >0\< C99 standard mandates minus sign but C89 does not ok 236 - >1e+04\< >1e+004\< three-digit exponent accepted ok 237 - >1.e+04\< >1.e+004\< three-digit exponent accepted ok 238 - >1.2e+04\< >1.2e+004\< three-digit exponent accepted ok 239 - >1.2e+04\< >1.2e+004\< three-digit exponent accepted ok 240 - >12345.6789\< ok 241 - > 12345.6789\< ok 242 - >0012345.6789\< ok 243 - >12345.6789 \< ok 244 - >12345.6789 \< ok 245 - >12345.6789 \< ok 246 - >-12345.7\< ok 247 - >-12345.7\< ok 248 - >1.23457e+06\< >1.23457e+006\< three-digit exponent accepted ok 249 - >+1.23457e+06\< >+1.23457e+006\< three-digit exponent accepted ok 250 - >1.23457e+06\< >1.23457e+006\< three-digit exponent accepted ok 251 - >-1.23457e+06\< >-1.23457e+006\< three-digit exponent accepted ok 252 - >-1.23457e+06\< >-1.23457e+006\< three-digit exponent accepted ok 253 - >-1.23457e+06\< >-1.23457e+006\< three-digit exponent accepted ok 254 - >0.00012345\< ok 255 - >1.2345e-05\< >1.2345e-005\< three-digit exponent accepted ok 256 - >1.23457e+102\< ok 257 - >1.23457e-102\< ok 258 - >0\< ok 259 - > 1.23457e+06\< > 1.23457e+006\< three-digit exponent accepted ok 260 - > +1.23457e+06\< >+1.23457e+006\< three-digit exponent accepted ******cut****** ok 544 - >30:31:32\< perl \#83194: vector flag + custom separator ok 545 - >030.031.032\< perl \#83194: vector flag + static precision ok 546 - >030.031.032\< perl \#83194: vector flag + dynamic precision ok 547 - >030:031:032\< perl \#83194: vector flag + custom separator + static precision ok 548 - >030:031:032\< perl \#83194: vector flag + custom separator + dynamic precision ok 549 - >118.101.114.115.105.111.110\< perl \#102586: vector flag + "version" ok 550 - > 011/ 022/ 033\< four reordered args Failed 1/550 subtests
Test Summary Report
op/sprintf.t (Wstat: 0 Tests: 550 Failed: 1) Failed test: 235 Files=1\, Tests=550\, 1 wallclock secs ( 0.08 usr + 0.00 sys = 0.08 CPU) Result: FAIL
-- bulk88 ~ bulk88 at hotmail.com
On Mon Jul 25 16:50:22 2016\, bulk88 wrote:
It wasn't fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d
I agree with bulk88. These lines handle the skip logic for plain win32:
140 # >comment skip: netbsd:vax-netbsd\< 141 $skip = $cond eq $archname; 142 }
That test has a skip condition "MSWin32"\, but my windows perls have archnames like "archname=MSWin32-x64-multi-thread". I suggest that any fix regex match `$archname =~ /\b$cond\b/` or `$archname =~ /^$cond/` rather than relying on string equality.
Copying/assigning to Jarkko\, as he committed 6151d43363
-- Respectfully\, Dan Collins
On Mon Jul 25 17:07:03 2016\, dcollinsn@gmail.com wrote:
On Mon Jul 25 16:50:22 2016\, bulk88 wrote:
It wasn't fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d
I agree with bulk88. These lines handle the skip logic for plain win32:
140 # >comment skip: netbsd:vax-netbsd\< 141 $skip = $cond eq $archname; 142 }
That test has a skip condition "MSWin32"\, but my windows perls have archnames like "archname=MSWin32-x64-multi-thread". I suggest that any fix regex match `$archname =~ /\b$cond\b/` or `$archname =~ /^$cond/` rather than relying on string equality.
Copying/assigning to Jarkko\, as he committed 6151d43363
Before tweaking any further I would like to understand how the mswin32 case matched before the 15899733e.
Before\, it used $^O instead of $archname. Come to think of it\, that is also probably fine - though I don't know if there are any unusual ways to build perl on Win32 that might end up with a strange $^O...
W:\buildbot\windows1\win64-mingww64-4_9_3\build>.\perl -Ilib -MConfig -wle "print $^O; print $Config{archname}" MSWin32 MSWin32-x64-multi-thread
On Mon\, Jul 25\, 2016 at 8:54 PM\, Jarkko Hietaniemi via RT \< perlbug-followup@perl.org> wrote:
On Mon Jul 25 17:07:03 2016\, dcollinsn@gmail.com wrote:
On Mon Jul 25 16:50:22 2016\, bulk88 wrote:
It wasn't fixed by 6151d433633f874a63f2a659b8bdaf400afacf4d
I agree with bulk88. These lines handle the skip logic for plain win32:
140 # >comment skip: netbsd:vax-netbsd\< 141 $skip = $cond eq $archname; 142 }
That test has a skip condition "MSWin32"\, but my windows perls have archnames like "archname=MSWin32-x64-multi-thread". I suggest that any fix regex match `$archname =~ /\b$cond\b/` or `$archname =~ /^$cond/` rather than relying on string equality.
Copying/assigning to Jarkko\, as he committed 6151d43363
Before tweaking any further I would like to understand how the mswin32 case matched before the 15899733e.
--- via perlbug: queue: perl5 status: open https://rt-archive.perl.org/perl5/Ticket/Display.html?id=128630
Does dbdc20d77 fix this?
--
Father Chrysostomos
On Fri Jul 29 00:27:33 2016\, sprout wrote:
Does dbdc20d77 fix this?
Probably that commit fixed it. Blead at Revision: c7202deeaccd41c87be26607b12888eb48962016 Author: Karl Williamson \khw@​cpan\.org Date: 7/31/2016 5:34:47 AM Message: locale.c: Add some DEBUG statements
passes in sprintf.t for me now so I think this ticket can be closed.
-- bulk88 ~ bulk88 at hotmail.com
Closed by OP request -- Karl Williamson
@khwilliamson - Status changed from 'open' to 'resolved'
Migrated from rt.perl.org#128630 (status was 'resolved')
Searchable as RT128630$