Re: pg_get_viewdefs() indentation considered harmful

I wrote:
> I'm still dubious about halving the step distance, because there are
> assorted places that adjust the indentation of specific keywords by
> distances that aren't a multiple of 2 (look for odd last arguments to
> appendContextKeyword).  I'm not sure how that will look after we make such
> a change.  Possibly it would work to only apply the scale factor to
> context->indentLevel, and not the indentPlus number?

I pushed a patch that does it that way, and also patches for the other
items discussed in this thread.
        regards, tom lane