[Dbix-class] Re: SQL::Translator conversion from Class::Base to Moo

[Peter Rabbitson]

On Sun, Sep 23, 2012 at 06:28:37PM +0100, Dagfinn Ilmari Mannsåker wrote:
> Peter Rabbitson <rabbit+dbic at rabbit.us> writes:
> 
> > On Sun, Sep 23, 2012 at 09:37:43AM -0500, fREW Schmidt wrote:
> >> As a followup to this, as things seem to have cooked sufficiently I have
> >> cut a minor fix release (0.11013) and rebased ilmari's changes on that
> >> and cut a dev release (0.11013_01.)  Assuming there aren't any issues
> >> from various smokers we'll probably cut a real release some time next week.
> >
> > Please review 7 janitorial commits and cut another rel: topic/mooify_cleanup
> 
> I disagree with the removal of the version from the 'use Moo'
> statements, since that changes a clear error message to a subtle bug.
> Just checking at build/test time is insufficient, since an older version
> of Moo might sneak in after SQLT was installed.
> 
> SQLT doesn't use Dist::Zilla or any other automatic prereq detection, so
> your argument against it is moot.
> 

Hence why it was left as the last commit, for easy throwaway.

If an older version of a prereq sneaks in after it was installed, you more 
or less have a broken perl install anyway, since you have no idea how much 
other stuff is affected. So in practitcal terms your argument is just as 
moot :) Furthermore spreading a ton of version declarations all over the 
place makes maintenance more problematic (not everyone is comfortable with 
perl -p -i -e), and hence invites the use of nuclear-powered packaging 
helpers. It's one step from there to dzillification.

Anyhow, if the commit doesn't make it in - it doesn't make it. The codebase 
maintainability suffers because of the explicit versioning, but it has 
enough other crap in it already anyway.

Cheers