Thread

1. Update hstore % Doc

   David E. Wheeler <david.wheeler@pgexperts.com> — 2010-08-06T20:45:46Z

   Hackers,
   
   I noticed that the hstore docs still document the => operator instead of %. This patch changes that. It also updates the first examples to us full SQL statements, because otherwise the use of => without surrounding single quotes was confusing.
   
   Best,
   
   David
   
   

2. Re: Update hstore % Doc

   Tom Lane <tgl@sss.pgh.pa.us> — 2010-08-06T22:16:11Z

   "David E. Wheeler" <david.wheeler@pgexperts.com> writes:
   > I noticed that the hstore docs still document the => operator instead
   > of %. This patch changes that.
   
   It looks to me like you are changing the examples of the I/O
   representation ... which did NOT change.
   
   			regards, tom lane
   
   

3. Re: Update hstore % Doc

   David E. Wheeler <david.wheeler@pgexperts.com> — 2010-08-06T22:21:10Z

   On Aug 6, 2010, at 3:16 PM, Tom Lane wrote:
   
   >> I noticed that the hstore docs still document the => operator instead
   >> of %. This patch changes that.
   > 
   > It looks to me like you are changing the examples of the I/O
   > representation ... which did NOT change.
   
   Hrm? The first few examples at the top? I find them confusing because there are no single quotes around them, so they look like the use of the deprecated => operator (especially the first two). Just look at: 
   
    http://developer.postgresql.org/pgdocs/postgres/hstore.html
   
   Maybe that's standard in the docs, but it seems weird to me. It's a minor point, though.
   
   We definitely need to document the `text % text` constructor rather than the deprecated `text => text` constructor.
   
   And I hate to say it, but % is awful. Sorry, I know I'm probably opening a can of worms here, and I did finally push % after all the arguments, but coming back to it fresh it just looks bizarre to me. Maybe that ship has sailed, though, and I'm just being difficult.
   
   Best,
   
   David
   
   
   

4. Re: Update hstore % Doc

   Tom Lane <tgl@sss.pgh.pa.us> — 2010-08-06T22:32:47Z

   "David E. Wheeler" <david.wheeler@pgexperts.com> writes:
   > On Aug 6, 2010, at 3:16 PM, Tom Lane wrote:
   >> It looks to me like you are changing the examples of the I/O
   >> representation ... which did NOT change.
   
   > Hrm? The first few examples at the top? I find them confusing because there are no single quotes around them, so they look like the use of the deprecated => operator (especially the first two). Just look at: 
   
   Yeah, but there's a sentence in front of them that says specifically
   that these are examples of the text representation, not pieces of SQL.
   
   > And I hate to say it, but % is awful.
   
   Yeah, I know, but you can't have =>.  Unless you can persuade the SQL
   committee to back off their syntax choice for parameters.  (:= would
   have been a lot better ...)
   
   			regards, tom lane
   
   

5. Re: Update hstore % Doc

   Tom Lane <tgl@sss.pgh.pa.us> — 2010-08-06T22:38:22Z

   "David E. Wheeler" <david.wheeler@pgexperts.com> writes:
   > We definitely need to document the `text % text` constructor
   
   BTW, there isn't any % constructor anymore --- we agreed to provide
   only the hstore(text, text) constructor.
   
   			regards, tom lane
   
   

6. Re: Update hstore % Doc

   David E. Wheeler <david.wheeler@pgexperts.com> — 2010-08-06T22:49:58Z

   On Aug 6, 2010, at 3:38 PM, Tom Lane wrote:
   
   >> We definitely need to document the `text % text` constructor
   > 
   > BTW, there isn't any % constructor anymore --- we agreed to provide
   > only the hstore(text, text) constructor.
   
   Oh, I must've been looking at an older checkout, then. Never mind.
   
   Best,
   
   David
   
   

7. Re: Update hstore % Doc

   Robert Haas <robertmhaas@gmail.com> — 2010-08-06T23:29:34Z

   On Fri, Aug 6, 2010 at 6:49 PM, David E. Wheeler
   <david.wheeler@pgexperts.com> wrote:
   > On Aug 6, 2010, at 3:38 PM, Tom Lane wrote:
   >
   >>> We definitely need to document the `text % text` constructor
   >>
   >> BTW, there isn't any % constructor anymore --- we agreed to provide
   >> only the hstore(text, text) constructor.
   >
   > Oh, I must've been looking at an older checkout, then. Never mind.
   
   And also - it was never the constructor.  It was briefly the slice operator.
   
   -- 
   Robert Haas
   EnterpriseDB: http://www.enterprisedb.com
   The Enterprise Postgres Company