Closed p5pRT closed 22 years ago
Bug and meta-Bug report. Bug report is that this devel perl fails tests in lib/peek.t\, notably:
$ ./perl t/lib/peek.t 1..17 ok 1 ok 2 ok 3 ok 4 ok 5 [SV = NV(0x8115c80) at 0x8129a4c REFCNT = 1 FLAGS = (PADTMP\,NOK\,pNOK) NV = -4.94660802946209209e+173 ] vs [SV = NV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(PADTMP\,NOK\,pNOK\) NV = 456] not ok 6 [SV = PVNV(0x813adb8) at 0x810644c REFCNT = 1 FLAGS = (NOK\,pNOK) IV = 0 NV = -5.41598767187353829e+25 PV = 0x810d140 "789"\0 CUR = 3 LEN = 4 ] vs [SV = PVNV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(NOK\,pNOK\) IV = 0 NV = 789\.1 PV = 0x[[:xdigit:]]+ "789"\\0 CUR = 3 LEN = 4] not ok 7 ok 8 ok 9 ok 10 [SV = RV(0x8139410) at 0x8141734 REFCNT = 1 FLAGS = (ROK) RV = 0x814171c SV = PVAV(0x813db04) at 0x814171c REFCNT = 2 FLAGS = () IV = 0 NV = 0 ARRAY = 0x8111718 FILL = 1 MAX = 1 ARYLEN = 0x0 FLAGS = (REAL) Elt No. 0 SV = IV(0x8110dec) at 0x8141674 REFCNT = 1 FLAGS = (IOK\,pIOK) IV = 123 Elt No. 1 SV = PVNV(0x813adf8) at 0x810ea58 REFCNT = 1 FLAGS = (IOK\,NOK\,pIOK\,pNOK) IV = 456 NV = -4.94660802946209209e+173 PV = 0 ] vs [SV = RV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(ROK\) RV = 0x[[:xdigit:]]+ SV = PVAV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 2 FLAGS = \(\) IV = 0 NV = 0 ARRAY = 0x[[:xdigit:]]+ FILL = 1 MAX = 1 ARYLEN = 0x0 FLAGS = \(REAL\) Elt No. 0 SV = IV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(IOK\,pIOK\) IV = 123 Elt No. 1 SV = PVNV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(IOK\,NOK\,pIOK\,pNOK\) IV = 456 NV = 456 PV = 0] not ok 11 [SV = RV(0x8139410) at 0x810ea58 REFCNT = 1 FLAGS = (ROK) RV = 0x814171c SV = PVHV(0x8113e50) at 0x814171c REFCNT = 2 FLAGS = (SHAREKEYS) IV = 1 NV = 0 ARRAY = 0x810d170 (0:7\, 1:1) hash quality = -0.0% KEYS = 1 FILL = 1 MAX = 7 RITER = -1 EITER = 0x0 Elt "123" HASH = 0xddce SV = PVNV(0x813adf8) at 0x8141674 REFCNT = 1 FLAGS = (IOK\,NOK\,pIOK\,pNOK) IV = 456 NV = -4.94660802946209209e+173 PV = 0 ] vs [SV = RV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(ROK\) RV = 0x[[:xdigit:]]+ SV = PVHV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 2 FLAGS = \(SHAREKEYS\) IV = 1 NV = 0 ARRAY = 0x[[:xdigit:]]+ \(0:7\, 1:1\) hash quality = 150.0% KEYS = 1 FILL = 1 MAX = 7 RITER = -1 EITER = 0x0 Elt "123" HASH = 0x[[:xdigit:]]+ SV = PVNV\(0x[[:xdigit:]]+\) at 0x[[:xdigit:]]+ REFCNT = 1 FLAGS = \(IOK\,NOK\,pIOK\,pNOK\) IV = 456 NV = 456 PV = 0] not ok 12 ok 13 ok 14 ok 15 ok 16 ok 17
Meta-bug report is that I'm confused. On this machine (very boring x86 clone) I can build 18 variations of perl without even touching threading or binary compatibility:
stdio/PerIO atop stdio/PerlIO atop sfio cross normal ints/64 bit ints/64 bit max cross doubles/long doubles
Would it be possible to set up some sort of autobuilder to exercise all the permutations?
The reason I ask is that I'm not quite sure which out of 64 bit ints\, 64 bit max\, long doubles or sfio is triggering the above problem. I've now got 6 builds of the same development perl to try and make the matrix of pass/fail to try to work where the failure is. And I feel no nearer. Hence bug report rather than patch.
Nicholas Clark
@jhi - Status changed from 'open' to 'resolved'
Call me callous but I'm closing off some Not OKs from 5.7.0 and 5.7.1.
@jhi - Status changed from 'open' to 'resolved'
Migrated from rt.perl.org#4502 (status was 'resolved')
Searchable as RT4502$