CPAN testing

cog on 2004-07-08T09:22:30

The latest release of Lingua::PT::Inflect dates from 20 May 2004... during this time, the module passed 3 tests (I'm talking about cpan-testers).

Two days ago, however, I received a report that it had failed on a test... the report said the module wouldn't work without Lingua::PT::Hyphenate, and it doesn't, but it also said I should add a line to Makefile.PL stating that (making use of the PREREQ_PM clause)... the problem is... that line is already there!!!

Am I missing something? Is this a bug in somebody else's code? :-| Any pointers, anyone? I'd be much appreciated :-|


CPANPLUS bug

LTjake on 2004-07-08T11:34:53

This is a known, but elusive, bug in CPANPLUS.

I've already created a ticket for it. I've just added information from your post as proof that it persists in version 0.049.

Known Problem

barbie on 2004-07-08T11:36:48

This has been a bug in CPANPLUS for a while, and was logged in RT 2 months ago. I believe Kane and Autrijus will be looking at this for the next release. Although this may appear to be a 0.48 issue, unfortunately there are additional issues with 0.49 that can prevent CPAN Testing, so upgrading is not always possible. However, I note that the report you refer to is tested with 0.49, so probably it is a bigger issue.

Personally I try and make sure that the reports are correct and that false reports are not generated by this issue with CPANPLUS. I currently have a backlog of modules I want to test, but can't because of this problem. Hopefully when Kane and Autrijus get the chance to work through this they will find the source of the problem. I've tried looking myself and it's not easy. CPANPLUS is a very complex piece of code.

So...

cog on 2004-07-08T11:50:03

Ok, now I understand what's going on... Thanks, LTjake and barbie.

Meanwhile, I take it as there's probably nothing to do about that failed test,
is there? I mean, the page for that module will continue to state that it
fails, when it doesn't... am I correct? :-|

Is the problem kind of random? Or are just the machines testing modules
randomly selected? :-) I suppose there is some chance that a new release may
or may not fail... right?

Re:So...

LTjake on 2004-07-08T12:31:45

Since no one has been able to track down the problem, it seems justified to call it "random."

My own personal testing machine (cpansmoke at alternation dot net - i586-linux 2.4.22-4tr) tests all incoming distributions. Some people only test certain modules. Some might use CPANPLUS to install modules and have reporting turned on -- thus creating tester reports. So, i guess it varies quite a bit.

As far as I can tell, once a report is sent, it's written in stone. But that one failed test is not something I'd worry about -- I'd be more concerned if it failed because of something I (as an author) had done! :)