Re: [Dbix-class] Set a resultset_class for a result_class from within a component loaded by the result_class

[David Ihnen]

Rob Kinyon wrote: On Thu, Mar 5, 2009 at 12:07, David Ihnen <dav...@norchemlab.com> wrote: Nothing is actually a plugin that impliments the 'Nothing' design pattern - that is - instead of returning null when referencing a row in an empty resultset (particularly ->single and ->first, which I use many times when NOT looping over results, hence resulting in undef dereference errors on empty) it instead returns a magical 'Nothing' object which has characteristics like - (->is_nothing === 1, ->id === 0, @list = $rs->method === (), $row = $rs->method === bless({}, Nothing).  The result - no need to if-then on empty resultsets when grabbing data. So, the following are different: my $x = $rs->method; my ($y) = $rs->method; That's sucky. interesting point.  If I had the Want module, I could easily return nothings if you have a specified number expected (1 in this case), or empty list if you expect infinite - since ($y) = $rs->method is different contextually than @y = $rs->method - which would make that not sucky. I wanted to use the Want module to change the return considering the context called in, but that module is, sadly, not compatible with mod_perl.  If it was, I could give back undefined, blessed ref, array ref, or hash ref depending on its context, which would be cool.  Dang segfaults... Context-awareness, while cool, is kinda like operator overloading and monkeypatching. Yes, it is. It's something you should have in that section of your toolbox that has the "Write 500 words as to why you should be allowed to use me" essay requirement. Well, in this case in order to cleanly express 'Nothing', which has different meanings in different contexts, I need to be able to detect that context so that I can be different for those contexts.  Within the limits of this particular design pattern, I think it is justifiable.   So though it is a bit sucky and constrains the implimentation patterns (my $y = $rs->method SHOULD be equal to my ($y) = $rs->method), but in my case the former is what you want) without the Want module... I think that the pattern has good benefits in being able to cut down on the if-then hooey.  How I hate if-then/tertiaries/logical operators/other repetitive code just to see if there's something there to operate on. At least with AUTOLOAD handling miscellaneous calls, I don't have to create special nothing hard-types one-to-parent-class as I would in Java/C++ that behave appropriately.  A nothing row that returns a nothing date-object, etc. David

_______________________________________________
List: http://lists.scsys.co.uk/cgi-bin/mailman/listinfo/dbix-class
IRC: irc.perl.org#dbix-class
SVN: http://dev.catalyst.perl.org/repos/bast/DBIx-Class/
Searchable Archive: http://www.grokbase.com/group/dbix-class@lists.scsys.co.uk