[DBIx-Class-Devel] DBIx-Class-0.08240-TRIAL docs review

[Peter Rabbitson]

On Fri, Feb 15, 2013 at 02:17:13PM -0500, Brendan Byrd wrote:
> See the latest commit to the topic/constructor_rewrite branch.  Mostly,
> it's a lot of commas.

I've never been good with these ;)

> *If L<collapsing|/collapse> is requested, there is an additional
> requirement of **selecting enough data to make every individual object
> uniquely identifiable.*
> 
> I thought collapse => 1 on prefetch.  Isn't it ALWAYS requested?

But that's the point of collapse - you can use it standalone. Consider
the bunch of "empty intermediates" here:
https://metacpan.org/source/RIBASUSHI/DBIx-Class-0.08240-TRIAL/t/prefetch/manual.t#L227

I also tried to explain this here, but if you missed it perhaps it needs 
more work:
https://metacpan.org/module/RIBASUSHI/DBIx-Class-0.08240-TRIAL/lib/DBIx/Class/ResultSet.pm#prefetch

> DBIC, DBIx::Class, or L<DBIx::Class>?  What is "standard"?  I'm fine with
> simply DBIC (after an introduction to the term).  The link doesn't really
> serve much of a purpose in this case.

No standard, a standard would certainly be good. Note also that the 
POD in question was written about a year ago originally. I adjusted it
to match a lot of the style decisions since, but missed the DBIC part.

On the other hand if we are converging on something, L<DBIC|DBIx::Class>
is probably the best way forward.

> *That cmp_ok() may or may not pass depending on the datasets involved. This
> behavior may or may not survive the 0.09 transition.*
> 
> Didn't you just fix this with this patch?  Or rather, if collapse=>1 was
> specified on the join, would that fix this?

The WHERE is something that acts on the resultset before it is even
returned to you. But the "may not survive 0.09" is crap, I will take
it out - there is no way we can break something so fundamental (even
if a bit coutnerintuitive).