Skip Menu |

This queue is for tickets about the Crypt-DSA CPAN distribution.

Report information
The Basics
Id: 13862
Status: new
Priority: 0/
Queue: Crypt-DSA

People
Owner: Nobody in particular
Requestors: joed [...] itsa.ucsf.edu
Cc:
AdminCc:

Bug Information
Severity: Critical
Broken in: 0.13
Fixed in: (no value)



Subject: getting error
Hi, I'm trying to get Crypt-DSA-0.13 working as part of the SSH-1.28 module on my AIX 5.1 system. I'm running perl 5.8.7 and have the C for AIX C compiler, and get a weird message. When itgets to the t/02-sign test, first it takes about 15 minutes before I get any output, and then I get the "Use of initialized value" error until I kill the process. See below. Should I try going back to 0.12? Joe DeBattista joed@itsa.ucsf.edu # make test PERL_DL_NONLAZY=1 /usr/local/bin/perl "-MExtUtils::Command::MM" "-e" "test_harness(0, 'inc', 'blib/lib', 'blib/arch')" t/00-compile.t t/01-util.t t/02-sign.t t/03-keygen.t t/04-pem.t t/06-fips.t t/07-openid.t t/00-compile....ok t/01-util.......ok t/02-sign....... Use of uninitialized value in numeric eq (==) at /home/root/.cpan/build/Crypt-DSA-0.13/blib/lib/Crypt/DSA/KeyChain.pm line 128. Use of uninitialized value in numeric eq (==) at /home/root/.cpan/build/Crypt-DSA-0.13/blib/lib/Crypt/DSA/KeyChain.pm line 128. Use of uninitialized value in numeric eq (==) at /home/root/.cpan/build/Crypt-DSA-0.13/blib/lib/Crypt/DSA/KeyChain.pm line 128. ...
From: Jurgen Pletinckx
Show quoted text
> Hi, > I'm trying to get Crypt-DSA-0.13 working as part of the SSH-1.28 > module on my AIX 5.1 system. I'm running perl 5.8.7 and have the
C Show quoted text
> for AIX C compiler, and get a weird message. When itgets to the > t/02-sign test, first it takes about 15 minutes before I get any > output, and then I get the "Use of initialized value" error until
I Show quoted text
> kill the process. See below. Should I try going back to 0.12? > > Joe DeBattista > joed@itsa.ucsf.edu > > # make test > PERL_DL_NONLAZY=1 /usr/local/bin/perl "-MExtUtils::Command::MM" "-e" > "test_harness(0, 'inc', 'blib/lib', 'blib/arch')" t/00-compile.t > t/01-util.t t/02-sign.t t/03-keygen.t t/04-pem.t t/06-fips.t t/07- > openid.t > t/00-compile....ok > t/01-util.......ok > t/02-sign....... > Use of uninitialized value in numeric eq (==) at > /home/root/.cpan/build/Crypt-DSA- > 0.13/blib/lib/Crypt/DSA/KeyChain.pm line 128.
Similar problem here; installing Crypt-DSA-0.13 on perl5.8.7 on Irix 6.5. A first problem in Irix occurs in Crypt/DSA/KeyChain.pm : KeyChain.pm:31: my $openssl = `which openssl`; KeyChain.pm:32: chomp $openssl; KeyChain.pm:33: if ($openssl) { On the typical Linux system, which will send the error message to STDERR if its argument is not found. On Irix 6.5, (and possibly on AIX), $openssl contains 'openssl not in /usr/sbin /usr/bsd /sbin /usr/bin /etc /usr/etc ' (truncated). I patched line 33 to if ($openssl and $openssl !~ /^openssl not in/) That will get you to the pureperl implementation, which is unfortunately very slow. Still, after half an hour, this gets me past generate_params and into generate_keys. Where I am caught in an apparently-infinite loop (again around lines 125-128). Impatient tester? Problem with Math::Bigint on Irix? Slow machine? Who can tell? Jurgen.pletinckx@algonomics.com
From: Jurgen Pletinckx
Show quoted text
> Still, after half an hour, this gets me past generate_params > and into generate_keys. > > Where I am caught in an apparently-infinite loop (again around lines > 125-128).
Following up on myself... This is an effect of the issue described in ticket 14281 "makerandom() not portable". So it's version 0.12 for me. Jurgen.pletinckx@algonomics.com
From: IVORW [...] cpan.org
[guest - Thu Sep 29 07:48:12 2005]: Trying on Solaris, I get a similar problem, but manifesting itself slightly differently, as the shell gives a different diagnostic when it can't find openssl. williami@otcdv2:/home/williami/CPANBLD/Crypt/DSA: make test PERL_DL_NONLAZY=1 /usr/local/bin/perl "-MExtUtils::Command::MM" "- e" "test_harness(0, 'inc', 'blib/lib', 'blib/arch')" t/00-compile.t t/01-util.t t/02-sign.t t/03-keygen.t t/04-pem.t t/06-fips.t t/07- openid.t t/00- compile....ok t/01- util.......ok t/02-sign.......sh: no: not found ... and hangs at this point. To my mind, the test should use findbin, rather than backticking which - not portable as there are different variants of 'which' in the Unix world.