Re: Recently added typedef "string" is a horrid idea

On Sun, Feb 7, 2016 at 8:03 AM, Peter Geoghegan <pg(at)heroku(dot)com> wrote:
> On Sat, Feb 6, 2016 at 2:11 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> Remember that the effects of typedef names are
>> *global*, so far as pgindent is concerned; not only varlena.c will
>> be affected.
>
> I'll remember that in the future.
>
>> Please rename this typedef with some less-generic name. Probably
>> some of the other identifiers added in the same commit should be
>> adjusted to match.
>
> I suggest "VarString". All the text SortSupport routines were renamed
> to match a pattern of "varstr.*" as part of the commit you mention.
>
> The implication that was intended by the rename is that the relevant
> routines are responsible for about the same cases as the cases handled
> by varstr_cmp(). I tend to mostly think of the text type when looking
> at varstr_cmp(), but it's also used by jsonb, for example, as well as
> char(n). It has a broader purpose; it is used by collatable types
> generally. So, a rename to "VarString" probably makes sense,
> independent of your pgindent concern.
>
> If this sounds like a good idea, I'll produce a patch soon.

VarString is OK with me - I'm not personally wedded to any specific
proposal here.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company