At 6:43 PM +0200 10/26/04, Abe Timmerman wrote: >Automated smoke report for 5.8.5 patch 23415 >FAFNER: VAX_7000-720 (VAX_7000-720/2 cpu) > on openvms - V7.2 > using CC/DECC version 60490005 > smoketime 5 hours 4 minutes (average 5 hours 4 minutes)
Congratulations, and thanks for getting this working. > >Summary: FAIL(F) > >O = OK F = Failure(s), extended report at the bottom >X = Failure(s) under TEST but not under harness >? = still running or test results not (yet) available >Build failures during: - = unknown or N/A >c = Configure, m = make, M = make (after miniperl), t = make test-prep > > 23415 Configuration (common) none >----------- --------------------------------------------------------- >F - >| +--------- -DDEBUGGING >+----------- no debugging > > >Failures: (common-args) none >[default] > [-.ext.B.t]optree_concise.t.............FAILED 12-13 16 > [-.ext.B.t]optree_sort.t................FAILED 3 5 7 9 > [-.ext.B.t]optree_specials.t............FAILED 2-6 > [-.ext.B.t]optree_varinit.t.............FAILED 6-8 13-15 20-22 These also fail on Alpha in maintperl, both with and without threads. They do not fail in a bleadperl built without threads. > [-.ext.Cwd.t]cwd.t......................FAILED 9-24 > [-.lib.ExtUtils.t]Install.t.............FAILED 17-29 > [-.lib.ExtUtils.t]basic.t...............FAILED 62 > [-.lib.ExtUtils.t]recurs.t..............FAILED 6 14 I'm not seeing these on Alpha and they are new to me. Details to vmsperl could be instructive. > [-.lib.Math.BigInt.t]bare_mbi.t.........FAILED 2766 > [-.lib.Math.BigInt.t]bigintpm.t.........FAILED 2766 > [-.lib.Math.BigInt.t]bigroot.t..........FAILED 1-8 > [-.lib.Math.BigInt.t]sub_mbi.t..........FAILED 2766 IIRC, Math::BigInt uses doubles to store chunks of big integers. The default floating point format on VAX is extremely limited in exponent range and probably breaks all sorts of assumptions. You might be able to get these to go away by configuring with -"Duser_c_flags=/FLOAT=G_FLOAT". I think I want to change the default to that for 5.10, but not in a maintenance release. The default for Alpha and Itanium is already to use IEEE floating point. > [-.lib]AutoSplit.t......................FAILED 2-3 5-58 New to me. > [-.lib]Carp.t...........................FAILED 10-11 16-19 Patch posted to p5p about an hour ago. > op/pack.t...............................FAILED ?? Probably also a floating point issue on VAX. >MANIFEST declared 'CHANGES5.8.1' but it is missing . . . >MANIFEST did not declare 'CHANGES5_8.1' Files with multiple dots in the name are a problem. Depending on what utility is used to unpack the archive and what switches are in effect, "foo.bar.baz" may turn into "foo_bar.baz", "foo.bar_baz", or "foo^.bar.baz". The manifest checker should handle all these cases, but I don't think even configure.com is doing the last one currently. >MANIFEST did not declare 'LIB/CONFIG.POD' >MANIFEST did not declare 'LIB/POD/PODCHECKER.COM' >MANIFEST did not declare 'LIB/POD/PODSELECT.COM' >MANIFEST did not declare 'PERL_SETUP.COM' >MANIFEST did not declare 'SMOKE58X.LOG' The realclean target needs a little bit of work, apparently. >MANIFEST did not declare 'T/DOIOWARN.TMP' >MANIFEST did not declare 'T/ERR' >MANIFEST did not declare 'T/TELL_WRITE.TXT' >MANIFEST did not declare 'T/TMON.OUT' And a few tests don't fully clean up after themselves. >-- >Report by Test::Smoke v1.19_51 build 737 running on perl 5.8.0 >(Reporter v0.018 / Smoker v0.018) -- ________________________________________ Craig A. Berry mailto:[EMAIL PROTECTED] "... getting out of a sonnet is much more difficult than getting in." Brad Leithauser
