On Fri, Feb 17, 2012 at 12:14 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> I don't know whether a similar improvement is
>> possible in this area, but we're certainly not going to get there by
>> labeling the user's expectations as unreasonable. I don't think they
>> are, and the people who wrote MySQL and Oracle evidently agree.
>
> The people who wrote MySQL had very poor taste in a lot of areas, and
> we are not going to blindly follow their lead. Oracle is not a terribly
> presentable system either. Having said that, I don't object to any
> clean improvements we can think of in this area --- but "make it work
> more like MySQL" had better not be the only argument for it.
Hey, if I preferred MySQL to PostgreSQL, I wouldn't be here. That
doesn't mean that there are exactly 0 things that they do better than
we do. What I'm unhappy about isn't that we're not bug-compatible
with MySQL, but rather that, in this case, I like MySQL's behavior
better, and the fact that they've made it work means it's not
theoretically impossible. It just involves some trade-off that I
don't believe we've thought about hard enough.
Standards compliance is a means to an end. The purpose of having
standards is to allow for interoperable implementations of the same
underlying functionality. That doesn't mean we should copy
non-standard warts, of course, but it isn't obvious to me that this is
a wart. No one has suggested that the user's actual query has more
than one reasonable interpretation, so complaining that it's ambiguous
doesn't impress me very much.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company