Bug #9369 for CPANPLUS: Prerelease Version Ordering

Mon Jan 03 11:56:57 2005 david [...] kineticode.com - Ticket created

CC:	Adrian Howard <adrianh [...] quietstars.com>
From:	David Wheeler <david [...] kineticode.com>
Subject:	Prerelease Version Ordering
Date:	Mon, 3 Jan 2005 08:55:17 -0800
To:	bug-cpanplus [...] rt.cpan.org

I'm not sure if this is an issue with CPANPLUS or with the way people think of versions, but CPANPLUS will not install a full version released after prerelease versions with the same minor version number. For example, I ran into this with Test::Class. I've been installing the prerelease versions for a while, which were numbered 0.06_01, 0.06_02, etc. When Adrian finally released the new version, it was 0.06. He was therefore thinking that the prerelease numbers are lower than 0.06 (although I recognize that, technically, they are not). In any event, CPANPLUS refused to install 0.06, telling me that I was up-to-date with one of the earlier pre-releases. Since, AFAIK, there is no documentation on whether one should put out prerelease versions between 0.05 and 0.06 that are numbered like 0.05_01 or 0.06_01, I suggest that CPANPLUS look at the dates (if it can, I realize that older versions aren't in the index) and install more recently released modules such as 0.06 when the previous releases have the same minor version number, such as 0.06_02. Or else someone should document what's a prerelease for what forthcoming version number and make sure everyone sees it! Thanks, David

Tue Jan 04 12:07:02 2005 david [...] kineticode.com - Correspondence added

CC:	Adrian Howard <adrianh [...] quietstars.com>
From:	David Wheeler <david [...] kineticode.com>
Subject:	Re: [cpan #9369] AutoReply: Prerelease Version Ordering
Date:	Tue, 4 Jan 2005 09:05:19 -0800
To:	bug-CPANPLUS [...] rt.cpan.org
RT-Send-Cc:

Copying Adrian's reply so that it goes into the same ticket. Show quoted text

> Hi David, > > On 3 Jan 2005, at 16:55, David Wheeler wrote: > [snip]

>> For example, I ran into this with Test::Class. I've been installing >> the prerelease versions for a while, which were numbered 0.06_01, >> 0.06_02, etc. When Adrian finally released the new version, it was >> 0.06. He was therefore thinking that the prerelease numbers are lower >> than 0.06 (although I recognize that, technically, they are not).

> > You're right that's the way I look at the ordering of prerelease > version numbers, although I can see that it's sane to argue it the > other way.

Indeed, if we consider things strictly numerically, it's the only thing that makes sense: % perl -le 'print 0.06 < 0.06_02' 1 So CPANPLUS is technically doing the right thing. I wonder if we shouldn't try to find a way to document this a the "correct" approach in general... Show quoted text

>> In any event, CPANPLUS refused to install 0.06, telling me that I was >> up-to-date with one of the earlier pre-releases.

> [snip] > > I threw out a 0.07 earlier today so you should get the newer version > from now on.

Thanks Adrian. Regards, David

Wed Jan 05 00:59:35 2005 adrianh [...] quietstars.com - Ticket #9418: Ticket created

CC:	bug-cpanplus [...] rt.cpan.org
From:	Adrian Howard <adrianh [...] quietstars.com>
Subject:	Re: Prerelease Version Ordering
Date:	Tue, 4 Jan 2005 13:16:56 +0000
To:	David Wheeler <david [...] kineticode.com>

Hi David, On 3 Jan 2005, at 16:55, David Wheeler wrote: [snip] Show quoted text

> For example, I ran into this with Test::Class. I've been installing > the prerelease versions for a while, which were numbered 0.06_01, > 0.06_02, etc. When Adrian finally released the new version, it was > 0.06. He was therefore thinking that the prerelease numbers are lower > than 0.06 (although I recognize that, technically, they are not).

You're right that's the way I look at the ordering of prerelease version numbers, although I can see that it's sane to argue it the other way. Show quoted text

> In any event, CPANPLUS refused to install 0.06, telling me that I was > up-to-date with one of the earlier pre-releases.

[snip] I threw out a 0.07 earlier today so you should get the newer version from now on. Cheers, Adrian

Mon Jan 10 10:01:59 2005 kane [...] cpan.org - Ticket #9418: Ticket 9418 MergedInto ticket 9369.

Mon Jan 10 10:38:13 2005 kane [...] cpan.org - Correspondence added

RT-Send-CC:

adrianh [...] quietstars.com

[adrianh@quietstars.com - Wed Jan 5 00:59:35 2005]: Show quoted text

> Hi David, > > On 3 Jan 2005, at 16:55, David Wheeler wrote: > [snip]

> > For example, I ran into this with Test::Class. I've been installing > > the prerelease versions for a while, which were numbered 0.06_01, > > 0.06_02, etc. When Adrian finally released the new version, it was > > 0.06. He was therefore thinking that the prerelease numbers are lower > > than 0.06 (although I recognize that, technically, they are not).

Not only are they 'technically' not, when comparing them using perl, we find that they are exactly the same: [kane@myriad ~]$ perl -le'print 1 if "0.06_02" == "0.06"' 1 And that is the way the $VERSION var is defined in each case =/ so there's no way for us to tell which one is higher or lower i'm afraid, until we use version objects everywhere Show quoted text

> You're right that's the way I look at the ordering of prerelease > version numbers, although I can see that it's sane to argue it the > other way. >

> > In any event, CPANPLUS refused to install 0.06, telling me that I was > > up-to-date with one of the earlier pre-releases.

> [snip]

--force will always save you here of course

Mon Jan 10 10:38:14 2005 kane [...] cpan.org - Status changed from 'new' to 'resolved'