All the Perl that's Practical to Extract and Report

sorry barbie - thanks for the good work

TeeJay on 2004-12-24T09:49:55

I just read barbie's journal entry about cpan-testers. Its pretty upsetting that somebody who has put in some good work in improving the standards of information and code in CPAN is driven from constructive work by whinging.

I have receieved a handful of fail reports and rt resports from cpan testers and rt.cpan.org. I get lots of bug reports from non-perl users for autodia so its water off a ducks back to me. But its valuable nonetheless.

Usually I will try and fix any problem straight away, but because cpan-testers is pretty automated I know that if any showstopper stops my code from installing or compiling it should usually get caught and reported back to me.

I'd really like to say thanks to barbie and the cpan-testers, they have ensured that my code is more likely to compile and install on places I either can't or just can't be bothered to to test it.

One to Many

Matts on 2004-12-24T17:39:29

Part of the problem is that it's one-to-many, so while Barbie gets lots of whinges back when CPANPLUS goes tits-up, the sender thinks he's just whinging once. I know I get frustrated by tester reports falsely accusing me of missing PREREQ_PM because it causes me to spend a period of time checking the code out of CVS and loading it into an editor, and then wondering if I've done something wrong. The trouble is clearly that many people complain to Barbie because he's such a prolific tester (and a very useful one too I might add) and gets bitten by this bug too often.

I don't think Barbie should quit though - he's a valuable asset. Hopefully he just needs some reassurance.

Moaning about cpan-testers

drhyde on 2004-12-25T00:15:05

I'll put my hand up and admit it. I've moaned at the CPAN testers. I'll continue doing it too. What I have moaned about is testers who submit their failure results without doing even minimal checking that their results are correct or, indeed, sane. Or about testers who don't even bother to read the documentation before submitting their failures. That is, testers who bizarrely think that quantity is better than quality!

Let's look at it from the other side of the fence. A few days ago I got a test failure for one of my modules. Never mind that the bug is actually in a *core* module that my module in turn uses (apparently some versions of Cwd are not taint-safe, maybe I should just not bother testing my code with -T so as not to tickle this), it's still my test failure. So I emailed the tester who'd found it within a matter of days with a suggested fix, and asked him to re-test on his system to see if my fix would work. I never heard anything back. If the whole point of the cpan-testers is to improve the quality of the CPAN, then I should have heard back from him. I'm still waiting. And it's not as if he's gone away on holiday or something like that, as he's submitted another (valid) test failure in that time!

Thanks Barbie

grantm on 2004-12-25T21:04:53

I also very much appreciate the efforts of Barbie and the other testers. The people who complain about a failure report implying there are bugs in their code are being bloody ridiculous.

Re:Thanks Barbie

drhyde on 2004-12-26T18:06:45

I agree. I'd never complain about that - I *welcome* genuine bug reports. However, I've had snotty auto-generated emails from some of the CPAN testers which were *not* to do with bugs, they were the result of stupid people blindly submitting their results without thinking.