> I don't care for the forced double quotes. Maybe if you double-quote
> only identifiers that need it?
I can do that. I did think about it, but it did seem to me that at a
glance, you can instantly pick out the names of the items from all the text,
rather than having to sort of search it. ie. it's consistent...
Ideally, I would have liked to have changed Index, Check Constraints and
Foreign Key Constraints into Constraints and Indexes, however I'm sure you
would prefer it if I made it clear that unique and primary constraints are,
in fact, indexes... I should suppress the 'btree' for primary and unique as
well since it's irrelevant.
> Upper-casing the keywords seems like a visual improvement.
Yep.
> I'm ambivalent about forcing the titles ("Indexes:" et al) onto their
> own lines. It's more readable, maybe, but it makes the display take up
> more vertical space. Having to scroll back negates any readability
> improvement IMHO...
Yes, but it improves horizontal utilization. Especially for long things
like Foreign Keys, it's annoying how they all start about 20 characters in
from the left - makes it quite hard to read...
Remember that most users in 7.3 shouldn't have all those triggers listed and
won't have rules. That will make it shorter in the general case...
> > Rules:
> > "test_rule" AS ON INSERT TO users_users DO INSTEAD NOTHING;,
> > "test_rule2" AS ON INSERT TO users_users DO INSTEAD NOTHING;
>
> The semicolon-comma bit is weird-looking, and confusing. Maybe we
> should lose the commas altogether.
yes, it is. The only way around it really is to mess about with not dumping
the last character of the pg_get_ruledef. Irritatingly enough,
pg_get_ruledef appends a semi-colon to its result, whereas all the others do
not...
Chris
