FAQ
Hi Gerald,

Debian is preparing to transition to perl 5.18, and test rebuilds of all
packages have revealed a problem with embperl (tracked at
http://bugs.debian.org/709668):
This package FTBFS with perl 5.18 from experimental (in a clean sbuild
session):

#51 escape.htm...
Error in Line 44
Is: >HashinA<ahref="http://localhost/tests?B=2&amp;A=1"><
Should: >HashinA<ahref="http://localhost/tests?A=1&amp;B=2"><

ERRORS detected! NOT all tests have been passed successfully
This is potentially due to the Hash overhaul / randomization in perl
5.18 as described at
http://search.cpan.org/dist/perl-5.18.0/pod/perldelta.pod#Hash_overhaul
and may require explicit sorting of keys where their order is relevant.

Will you have time to look into this?

best,
Florian


---------------------------------------------------------------------
To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
For additional commands, e-mail: embperl-help@perl.apache.org

Search Discussions

  • Gerald Richter at May 27, 2013 at 3:50 am
    Hi Florian,

    I will take a look at this issue during the next weekend

    Gerald

    -----Ursprüngliche Nachricht-----
    Von: Florian Schlichting
    Gesendet: Samstag, 25. Mai 2013 18:41
    An: embperl@perl.apache.org
    Cc: 709668@bugs.debian.org
    Betreff: embperl: test failure with perl 5.18

    Hi Gerald,

    Debian is preparing to transition to perl 5.18, and test rebuilds of all packages
    have revealed a problem with embperl (tracked at
    http://bugs.debian.org/709668):
    This package FTBFS with perl 5.18 from experimental (in a clean sbuild
    session):

    #51 escape.htm...
    Error in Line 44
    Is: >HashinA<ahref="http://localhost/tests?B=2&amp;A=1"><
    Should: >HashinA<ahref="http://localhost/tests?A=1&amp;B=2"><

    ERRORS detected! NOT all tests have been passed successfully
    This is potentially due to the Hash overhaul / randomization in perl
    5.18 as described at
    http://search.cpan.org/dist/perl-5.18.0/pod/perldelta.pod#Hash_overhaul
    and may require explicit sorting of keys where their order is relevant.

    Will you have time to look into this?

    best,
    Florian


    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org


    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org
  • Gerald Richter at Jul 16, 2013 at 4:45 am
    Hi Florian,

    hash randomization and other changes in Perl 5.18 did make things much more complicated than I expected.

    I have now a first version that seems to work with Perl 5.18 in the svn.

    I have to do some further tests, than I will create a new release

    Gerald
    -----Ursprüngliche Nachricht-----
    Von: Florian Schlichting
    Gesendet: Samstag, 25. Mai 2013 18:41
    An: embperl@perl.apache.org
    Cc: 709668@bugs.debian.org
    Betreff: embperl: test failure with perl 5.18

    Hi Gerald,

    Debian is preparing to transition to perl 5.18, and test rebuilds of all packages
    have revealed a problem with embperl (tracked at
    http://bugs.debian.org/709668):
    This package FTBFS with perl 5.18 from experimental (in a clean sbuild
    session):

    #51 escape.htm...
    Error in Line 44
    Is: >HashinA<ahref="http://localhost/tests?B=2&amp;A=1"><
    Should: >HashinA<ahref="http://localhost/tests?A=1&amp;B=2"><

    ERRORS detected! NOT all tests have been passed successfully
    This is potentially due to the Hash overhaul / randomization in perl
    5.18 as described at
    http://search.cpan.org/dist/perl-5.18.0/pod/perldelta.pod#Hash_overhaul
    and may require explicit sorting of keys where their order is relevant.

    Will you have time to look into this?

    best,
    Florian


    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org


    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org
  • Damyan Ivanov at Aug 26, 2013 at 6:09 am
    Hi, Gerald,

    -=| Gerald Richter - ECOS, 16.07.2013 06:44:11 +0200 |=-
    hash randomization and other changes in Perl 5.18 did make things
    much more complicated than I expected.

    I have now a first version that seems to work with Perl 5.18 in the svn.

    I have to do some further tests, than I will create a new release
    I tried to apply the patch from revision 1503556 to the current debian
    sources (version 2.5.0 rc3). The patch applied with only one rejected
    hunk, which I applied by hand.

    Embperl built fine and tests that previously failed (51 escape.htm)
    now pass. However, some other tests not fail.

    With perl 5.18:

      #32 notfound.htm...

      Expected 1 more error(s) in logfile

      Input: test/html/notfound.htm
      Output: test/tmp/out.htm
      Log: test/tmp/test.log
      Testparameter:
        errors = 1

    and with perl 5.14:

      #247 rtf/rtfadv.asc... Error in Line 52
      Is: >{Ulrike}<
      Should: >{Sarah}<

    If the tests pass for you, I guess we could try packaging a snapshot
    from SVN's trunk.
    (https://metacpan.org/module/GRICHTER/Embperl-2.4.0/SVN.pod)

    Best regards,
         Damyan,
         Debian Perl Group

    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org
  • Gerald Richter at Aug 26, 2013 at 6:14 am
    Hi Damyan and Florian,

    thanks for testing and the patches.

    I will take a closer look to it during the next few days, create a new tar.gz to make sure nothing is missing and give you a feedback.

    Gerald
    -----Ursprüngliche Nachricht-----
    Von: Damyan Ivanov
    Gesendet: Freitag, 23. August 2013 16:36
    An: Gerald Richter; 709668@bugs.debian.org
    Cc: Florian Schlichting; embperl@perl.apache.org
    Betreff: Re: Bug#709668: AW: embperl: test failure with perl 5.18

    Hi, Gerald,

    -=| Gerald Richter - ECOS, 16.07.2013 06:44:11 +0200 |=-
    hash randomization and other changes in Perl 5.18 did make things much
    more complicated than I expected.

    I have now a first version that seems to work with Perl 5.18 in the svn.

    I have to do some further tests, than I will create a new release
    I tried to apply the patch from revision 1503556 to the current debian sources
    (version 2.5.0 rc3). The patch applied with only one rejected hunk, which I
    applied by hand.

    Embperl built fine and tests that previously failed (51 escape.htm) now pass.
    However, some other tests not fail.

    With perl 5.18:

    #32 notfound.htm...

    Expected 1 more error(s) in logfile

    Input: test/html/notfound.htm
    Output: test/tmp/out.htm
    Log: test/tmp/test.log
    Testparameter:
    errors = 1

    and with perl 5.14:

    #247 rtf/rtfadv.asc... Error in Line 52
    Is: >{Ulrike}<
    Should: >{Sarah}<

    If the tests pass for you, I guess we could try packaging a snapshot from
    SVN's trunk.
    (https://metacpan.org/module/GRICHTER/Embperl-2.4.0/SVN.pod)

    Best regards,
    Damyan,
    Debian Perl Group

    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org


    ---------------------------------------------------------------------
    To unsubscribe, e-mail: embperl-unsubscribe@perl.apache.org
    For additional commands, e-mail: embperl-help@perl.apache.org
  • Axel Beckert at Jan 26, 2014 at 1:15 am
    Hi Gerald,
    On Mon, Aug 26, 2013 at 08:13:41AM +0200, Gerald Richter wrote:
    I will take a closer look to it during the next few days, create a
    new tar.gz to make sure nothing is missing and give you a feedback.
    I just noticed there's a new Embperl 2.5.0 RC4 and checked how far it
    comes with regards to the test suite.

    There are still a bunch of hash randomisation issues. I've attached a
    patch which fixes all of those I encountered so far by running the
    test suite many dozen times. I would expect that you could apply this
    patch to your code without changes.

    Further I disabled all tests which just check more or less Perl
    version specific error messages (i.e. includeerr{2,3}.htm) in our
    package build. (But I think that should be solved differently
    upstream, if it's an issue for you, too.)

    This leaves so far two issues which look like real bugs:

    #66 lists.htm fails occassionally and non-deterministically, but it
    doesn't seem to be a hash randomisation issue. If it fails, it looks
    like this:

    #66 lists.htm...
    Error in Line 54
    Is: ></select></p><
    Should: ><optionvalue="B"SELECTED>2</option><
    Input: test/html/lists.htm
    Output: test/tmp/out.htm
    Compared to: test/cmp/lists.htm
    Log: test/tmp/test.log
    Testparameter:
       query_info = sel=2&SEL1=B&SEL3=D&SEL4=cc

    And if I check test/tmp/out.htm vs test/cmp/lists.htm, I see

         <p><select name="D1" size="1">
             <option value="A" >1</option>
         </select></p>

    but should see

         <p><select name="D1" size="1">
             <option value="A" >1</option>

             <option value="B" SELECTED>2</option>

             <option value="C" >3</option>

             <option value="D" >4</option>

             <option value="E" >5</option>

             <option value="F" >6</option>
         </select></p>

    This does not look like a hash randomisation issue to me and hence may
    be a real bug.

    If #66 passes, the next one which fails (and always fails) is #147
    exit.htm:

    #147 exit.htm...
    Error in Line 11
    Is: After Exit
    Should: <EOF>
    Input: test/html/exit.htm
    Output: test/tmp/out.htm
    Compared to: test/cmp/exit.htm
    Log: test/tmp/test.log
    Testparameter:
       cgi = 0

    This looks like a real bug as "[- exit -]" seems to have no effect but
    should have an effect.

    Hope this helps.

       Regards, Axel
    --
      ,''`. | Axel Beckert <abe@debian.org>, http://people.debian.org/~abe/
    : :' : | Debian Developer, ftp.ch.debian.org Admin
    `. `' | 1024D: F067 EA27 26B9 C3FC 1486 202E C09E 1D89 9593 0EDE
       `- | 4096R: 2517 B724 C5F6 CA99 5329 6E61 2FF9 CD59 6126 16B5

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupembperl @
categoriesmodperl, perl
postedMay 27, '13 at 3:40a
activeJan 26, '14 at 1:15a
posts6
users4
websiteperl.apache.org

People

Translate

site design / logo © 2018 Grokbase