Bug #83369 for Crypt-Random-Seed: Load Crypt::Random::TESHA2 only after OS check in t/04-win32.t

Sat Feb 16 09:18:08 2013 CARNIL [...] cpan.org - Ticket created

Subject:	Load Crypt::Random::TESHA2 only after OS check in t/04-win32.t
Date:	Sat, 16 Feb 2013 15:17:54 +0100
To:	bug-Crypt-Random-Seed [...] rt.cpan.org
From:	Salvatore Bonaccorso <carnil [...] cpan.org>

Hi Dana I'm currently trying to package Crypt::Random::Seed (thanks!) and following that Bytes::Random::Secure for Debian. During first look I noticed in t/04-win32.t: Would it be possible to put the BEGIN block checking for Win32 before loading a module used in the test, specifically Crypt::Random::TESHA2 in this case. ----cut---------cut---------cut---------cut---------cut---------cut----- --- a/t/04-win32.t +++ b/t/04-win32.t @@ -1,7 +1,6 @@ #!/usr/bin/env perl use strict; use warnings; -use Crypt::Random::TESHA2; BEGIN { if ($^O ne 'MSWin32') { @@ -10,6 +9,8 @@ } } +use Crypt::Random::TESHA2; + use Test::More tests => 3; require_ok("Win32"); ----cut---------cut---------cut---------cut---------cut---------cut----- Many thanks for your development of this module! Regards, Salvatore

Sat Feb 16 12:38:48 2013 DANAJ [...] cpan.org - Taken

Sat Feb 16 12:53:46 2013 DANAJ [...] cpan.org - Correspondence added

RT-Send-CC:

carnil [...] cpan.org

Salvatore, thank you very much for the report. I just put version 0.03 on CPAN that should fix this. t/04-win32.t no longer loads TESHA2 at all, and t/13-whitelist.t will skip the test (with some diagnostic messages) if TESHA2 isn't installed. Is it possible that you're seeing this because the dependency Crypt::Random::TESHA2 isn't installed? In theory that shouldn't happen (Makefile.PL will indicate: "Warning: prerequisite Crypt::Random::TESHA2 0 not found."). It was the only way I could reproduce the issue on Debian systems. However, the whole point of using TESHA2 was to ensure we had a randomness source even for systems without a standard O/S randomness source. Debian should always have /dev/random and /dev/urandom, so there really is no need for TESHA2 on that system. Let me know if the new version fixes the issue! Thanks, Dana On Sat Feb 16 09:18:08 2013, CARNIL wrote: Show quoted text

> Hi Dana > > I'm currently trying to package Crypt::Random::Seed (thanks!) and > following that Bytes::Random::Secure for Debian. During first look I > noticed in t/04-win32.t: > > Would it be possible to put the BEGIN block checking for Win32 before > loading a module used in the test, specifically Crypt::Random::TESHA2 > in this case. > > ----cut---------cut---------cut---------cut---------cut---------cut----- > --- a/t/04-win32.t > +++ b/t/04-win32.t > @@ -1,7 +1,6 @@ > #!/usr/bin/env perl > use strict; > use warnings; > -use Crypt::Random::TESHA2; > > BEGIN { > if ($^O ne 'MSWin32') { > @@ -10,6 +9,8 @@ > } > } > > +use Crypt::Random::TESHA2; > + > use Test::More tests => 3; > > require_ok("Win32"); > ----cut---------cut---------cut---------cut---------cut---------cut----- > > Many thanks for your development of this module! > > Regards, > Salvatore >

Sat Feb 16 12:53:47 2013 The RT System itself - Status changed from 'new' to 'open'

Sat Feb 16 16:53:30 2013 CARNIL [...] cpan.org - Correspondence added

Subject:	Re: [rt.cpan.org #83369] Load Crypt::Random::TESHA2 only after OS check in t/04-win32.t
Date:	Sat, 16 Feb 2013 22:53:16 +0100
To:	Dana Jacobsen via RT <bug-Crypt-Random-Seed [...] rt.cpan.org>
From:	Salvatore Bonaccorso <carnil [...] cpan.org>

Hi Dana On Sat, Feb 16, 2013 at 12:53:46PM -0500, Dana Jacobsen via RT wrote: Show quoted text

> <URL: https://rt.cpan.org/Ticket/Display.html?id=83369 > > > Salvatore, thank you very much for the report. > > I just put version 0.03 on CPAN that should fix this. t/04-win32.t no > longer loads TESHA2 at all, and t/13-whitelist.t will skip the test > (with some diagnostic messages) if TESHA2 isn't installed. > > Is it possible that you're seeing this because the dependency > Crypt::Random::TESHA2 isn't installed? In theory that shouldn't happen > (Makefile.PL will indicate: "Warning: prerequisite Crypt::Random::TESHA2 > 0 not found."). It was the only way I could reproduce the issue on > Debian systems. > > However, the whole point of using TESHA2 was to ensure we had a > randomness source even for systems without a standard O/S randomness > source. Debian should always have /dev/random and /dev/urandom, so > there really is no need for TESHA2 on that system.

Yes this is absolutely true. I have not (yet) packaged Crypt::Random::TESHA2 and wanted to avoid the dependency on it, as Debian should always have /dev/random and /dev/urandom. This was exaclty why wanted to avoid to package TESHA2 in principle only for the tests. Thank you a lot for your promt presponse! Regards, Salvatore

Sat Feb 16 17:31:03 2013 DANAJ [...] cpan.org - TimeEstimated changed from (no value) to '30'

Sat Feb 16 17:31:03 2013 DANAJ [...] cpan.org - TimeWorked changed from (no value) to '30'

Sat Feb 16 17:31:03 2013 DANAJ [...] cpan.org - Status changed from 'open' to 'resolved'

Sat Feb 16 17:31:04 2013 DANAJ [...] cpan.org - Broken in 0.02 added

Sat Feb 16 17:31:04 2013 DANAJ [...] cpan.org - Severity Normal added

Sat Feb 16 17:31:04 2013 DANAJ [...] cpan.org - Fixed in 0.03 added