I'd always thought renaming the tarball to contain an underscore was an
acceptable way to mark a distribution as dev-only.

Chris is hypothesising that that may have confused CPANPLUS.

Shouldn't CPANPLUS be robust against such things, by untaring into an
empty directory, for example? Or else flag it as an error?

Meanwhile, I'll make another dev release which untars into a distinct
directory and we'll see how that works out in testing.

Tim.

----- Forwarded message from Chris 'BinGOs' Williams <[email protected]> 
-----

Date: Tue, 3 Feb 2009 11:54:28 +0000
From: Chris 'BinGOs' Williams <[email protected]>
To: Tim Bunce <[email protected]>
Subject: Re: (Fwd) Re: What's the common denominator in these NYTProf
        failures?

Hi, Tim,

Okay I had a chance to test manually on one of the vms where the tests were 
failing.

As soon as fetched T/TI/TIMB/Devel-NYTProf-2.07_94.tar.gz from a mirror and 
extracted it
I guessed what the problem might be.

That tarball extracted to Devel-NYTProf-2.07/

That set alarm bells ringing straight away in my mind.

Anyways, 'make test' was sucesssfull in that extracted directory.

So, I ran it through minismokebox and that was successfull as well.

What I think has been happening is that because the tarballs have the same 
extraction
directory as previous dev releases ( and the last stable release ), they get 
extracted over
each other and we see artifacts from the previous versions, hence the FAIL 
reports.

You don't see this with CPAN.pm as it always adds a random string onto the 
extraction directory.
CPANPLUS tends to trust that the CPAN author has done the right thing[tm].

Cheers,

-- 
Chris Williams
aka BinGOs
PGP ID 0x4658671F
http://www.gumbynet.org.uk
==========================



----- End forwarded message -----

Reply via email to