wrong default join type for config.rule_age_hold_protection in reports

And here's the branch: http://git.evergreen-ils.org/?p=working/Evergreen.git;a=shortlog;h=refs/heads/user/csharp/lp1419813_fix_crahp_joins_for_reporter

Pushed to master, rel_2_6, and rel_2_7. Thanks, Chris!

Following discussion on #evergreen (http://irc.evergreen-ils.org/evergreen/2015-02-23#i_158837), "might_have" doesn't end up producing the left join that is desired, so I've done as requested and pushed a revert to master, rel_2_6, and rel_2_7.

I'd like to add my apologies for insufficient testing of this patch before submitting it here. My cursory initial tests were fine, but building another template revealed this issue.

Thanks, Galen.

Chris

Well, if there is in fact a bit of egg on your face, there is on mine too.

Thanks again, Galen.

Well, moving onwards, though, I'd like to reframe this ticket to be broader than "the IDL has the wrong join type".

The problem at this point is that "has_a" is conceptually incorrect here because not every item has an entry in config.rule_age_hold_protect, but changing that to "might_have" causes the SQL builder (either in the JS or Perl layer) to join on crahp.id = asset.copy.id (or reporter.classic_item_list.id) which is obviously wrong and ignores the IDL.

The suggested workaround was to keep the "rel" type as "has_a" and use nullability to force the correct join. But when I tried that, it only allows "Parent" nullability (e.g., a RIGHT OUTER LOIN) when what's needed is a LEFT (OUTER) JOIN. So until this is fixed, there is not a way to create a report template that will display hold age protection rules correctly.

While this is the only instance of this issue I've seen, I would suspect the problem is more widespread.