Steve Crawford wrote:
> An example: I have asked on these groups several times about arrays and have
> gotten no replies. Currently if I ask for an element beyond the end of an
> array it will return a null. If I could find that to be the documented
> specification for array behavior I could start working on my app. As it is I
> can not because I may be creating code that will break with the next release
> of PostgreSQL.
>
> I am all the more concerned since with version 7.3 the interpretation of
> arrays has become non-consistent. A skipped element in a text array gets a
> default entry of '' while a skipped element in an int array, which used to
> get a default entry of 0, now causes an error. So before we can chastize
> someone for not following a specification there has to actually be one.
If you have an itch, then I'd suggest you scratch it ;-)
I'm sure if you submitted patches to improve the current documentation, they'd
be gratefully accepted.
Similarly, if there are currently unspecified aspects of array handling
behavior that concern you, post a proposal to HACKERS and work with the
community to refine and gain acceptance of that proposal. Even if you don't
actually do the coding, those first few steps (thinking through how arrays
*should* behave, clearly and completely *explaining* it, and getting community
*agreement*) are typically non-trivial, and a significant barrier for anyone
wanting to step up and write the code.
Just my 2c.
Joe