On 23 April 2015 at 14:50, Andres Freund <and...@anarazel.de> wrote:

> > ​Maybe I'm misreading it, but isn't index_predicate meant to be inside
> the
> > brackets?
> >
> >
> http://postgres-benchmarks.s3-website-us-east-1.amazonaws.com/on-conflict-docs/sql-insert.html
>
> That has changed since.


​Oh, helpful. :)​

​I'll shut up. I have a feeling that my objection is really with the very
idea of unreserved keywords and I have a feeling that there will be rather
more people shouting me down if I go off on that particular rant; meanwhile
it's 20 years since I used yacc in earnest and it's too hazy to be able to
argue about what it is or isn't capable of.

When I set out I was really only hoping to express a preference as a user;
on balance I would really rather not have DO IGNORE, if it were possible to
avoid, because it's really ugly, but DO UPDATE/DO NOTHING I could just
about cope with (and means you don't need to add IGNORE as a keyword,
win!), although it still mildly pains me that there's an additional
unnecessary word.

But I certainly don't object enough to hold up you guys doing the actual
work for my benefit (among others, obviously!).

G

Reply via email to