Re: [PATCH] Exorcise "zero-dimensional" arrays (Was: Re: Should array_length() Return NULL)

On Thu, Jun 20, 2013 at 07:13:48PM -0500, Merlin Moncure wrote:
> On Thu, Jun 20, 2013 at 6:40 PM, Bruce Momjian <bruce@momjian.us> wrote:
> >> Kinda -- what I'm saying is you just don't go around changing function
> >> behaviors to make them 'better' unless the affected behavior was
> >> specifically reserved as undefined.  The fact is nobody knows how many
> >> users will be affected and the extent of the ultimate damage (pro tip:
> >> it's always more and worse than expected); I'm astonished it's even
> >> being considered.
> >
> > Well, I think the question is how many people have such arrays that will
> > be effected.  If we don't do something, we live with this odd behavior
> > forever.  We have been willing to make some bold decisions in the past
> > to improve user experience, and it mostly has worked out well.  I
> > disagree that it is always worse than expected.
> 
> Well, you can have the last word (although 'bold' was an interesting
> word choice, heh)  -- I feel guilty enough about beating up Brendan
> already.  I feel this way every time compatibility changes come up, so
> it's nothing specific to this patch really.

Well, sometimes we underestimate the impact of changes, sometimes we
overestimate.  The big problem is weighing the short-term problems of
change but not the long-term benefit of a change.  This array problem
goes back to at least 2008:
http://www.postgresql.org/message-id/28026.1224611437@sss.pgh.pa.us

so we have at least five years of confusion by not changing it then.  I
am not saying we need to change it, but do think we need to weigh both
issues.

--  Bruce Momjian  <bruce@momjian.us>        http://momjian.us EnterpriseDB
http://enterprisedb.com
 + It's impossible for everything to be true. +