Skip Menu |

This queue is for tickets about the Perinci-CmdLine-Classic CPAN distribution.

Report information
The Basics
Id: 110234
Status: resolved
Priority: 0/
Queue: Perinci-CmdLine-Classic

People
Owner: Nobody in particular
Requestors: SREZIC [...] cpan.org
Cc:
AdminCc:

Bug Information
Severity: (no value)
Broken in:
  • 1.60
  • 1.61
  • 1.62
Fixed in: (no value)



Subject: Test failures
Here's the overview: http://matrix.cpantesters.org/?dist=Perinci-CmdLine-Classic%201.60 Sorry, no additional information yet.
On Wed Dec 09 01:24:16 2015, SREZIC wrote: Show quoted text
> Here's the overview: http://matrix.cpantesters.org/?dist=Perinci- > CmdLine-Classic%201.60 > > Sorry, no additional information yet.
Currently cannot reproduce, even after updating all modules ('cpan-outdated | cpanm -n').
On 2015-12-10 05:55:40, PERLANCAR wrote: Show quoted text
> On Wed Dec 09 01:24:16 2015, SREZIC wrote:
> > Here's the overview: http://matrix.cpantesters.org/?dist=Perinci- > > CmdLine-Classic%201.60 > > > > Sorry, no additional information yet.
> > Currently cannot reproduce, even after updating all modules ('cpan- > outdated | cpanm -n').
Possible reason: tests fail if Log::Dispatch::FileWriteRotate is not installed?
On 2015-12-12 03:06:13, SREZIC wrote: Show quoted text
> On 2015-12-10 05:55:40, PERLANCAR wrote:
> > On Wed Dec 09 01:24:16 2015, SREZIC wrote:
> > > Here's the overview: http://matrix.cpantesters.org/?dist=Perinci- > > > CmdLine-Classic%201.60 > > > > > > Sorry, no additional information yet.
> > > > Currently cannot reproduce, even after updating all modules ('cpan- > > outdated | cpanm -n').
> > Possible reason: tests fail if Log::Dispatch::FileWriteRotate is not > installed?
Still happens with 1.62.
From the fail reports for 1.61 (fails for 1.62 are not coming in yet), I see two remaining kinds: 1) "Died: Unknown arguments to new(): Filter, l4p_depends_on, l4p_post_config_subs, min_level, name at /usr/perl5.18.2/lib/site_perl/5.18.2/Log/Dispatch/FileWriteRotate.pm line 27." This is caused by an older version of Log::Dispatch::FileWriteRotate (< 0.03) which is a prereq for Log::Any::App, which is a prereq for Perinci::CmdLine::Classic. How do you propose resolving error reports like this? Should I release a new version of Log::Any::App just to bump the prereq version of Log::Dispatch::FileWriteRotate? Should I then also release a new version of Perinci::CmdLine::Classic just to bump the prereq version of Log::Any::App? For a large dependency tree, this could lead to loads of new minor releases. On the other hand, testers could test with a clean(er)/fresh(er) installation and purge old versions of modules. 2) "Record #0 of streaming argument 'input' fails validation: Not of type decimal number at (eval 189) line 50, <> line 1." I haven't pinned the cause yet, probably has something to do with older versions of Data::Sah or other modules, but this is not listed in the reports because Data::Sah is an indirect prerequisite. I am thinking of adding a routine in the test scripts to print out the content of %INC and the installed versions of loaded modules listed there, to help debugging the problem. What do you think of that? Do you know if something like this is already supported?
On 2016-01-14 15:44:28, PERLANCAR wrote: Show quoted text
> From the fail reports for 1.61 (fails for 1.62 are not coming in yet), > I see two remaining kinds: > > 1) "Died: Unknown arguments to new(): Filter, l4p_depends_on, > l4p_post_config_subs, min_level, name at > /usr/perl5.18.2/lib/site_perl/5.18.2/Log/Dispatch/FileWriteRotate.pm > line 27." > > This is caused by an older version of Log::Dispatch::FileWriteRotate > (< 0.03) which is a prereq for Log::Any::App, which is a prereq for > Perinci::CmdLine::Classic. > > How do you propose resolving error reports like this? Should I release > a new version of Log::Any::App just to bump the prereq version of > Log::Dispatch::FileWriteRotate? Should I then also release a new > version of Perinci::CmdLine::Classic just to bump the prereq version > of Log::Any::App? For a large dependency tree, this could lead to > loads of new minor releases.
In a perfect world, this would be the right solution. Show quoted text
> On the other hand, testers could test > with a clean(er)/fresh(er) installation and purge old versions of > modules.
This is not a good solution. Test machines should reflect the real world of real users. And there are different types of users: some who always install everything from scratch on an empty system; some who keep everything up-to-date; and some who only update or install modules if they have to, typically having a somewhat outdated system (the latter could be seen as a loosened "never touch a running system" approach). So a module author cannot assume the state of a user's installation to be completely up-to-date. Show quoted text
> 2) "Record #0 of streaming argument 'input' fails validation: Not of > type decimal number at (eval 189) line 50, <> line 1." I haven't > pinned the cause yet, probably has something to do with older versions > of Data::Sah or other modules, but this is not listed in the reports > because Data::Sah is an indirect prerequisite. > > I am thinking of adding a routine in the test scripts to print out the > content of %INC and the installed versions of loaded modules listed > there, to help debugging the problem. What do you think of that? Do > you know if something like this is already supported?
Well, then you're maybe interested in https://rt.cpan.org/Ticket/Display.html?id=109900 CPAN::Testers::ParseReport is used for creating the statistical reports at http://analysis.cpantesters.org/ , so using Module::Versions::Report (or a compatible output) could help to find problematic dependencies automatically.
Released some new versions to force update: - Perinci-CmdLine-Classic-1.63 depends on Log::Any::App 0.52 - Log-Any-App-0.52 depends on Log-Dispatch::FileWriteRotate 0.04 - Log-Dispatch-FileWriteRotate-0.04 depends on File::Write::Rotate 0.29 (>0.28)