Re: ts_rank error/result of '1e-020'

Gavin Flower <GavinFlower@xxxxxxxxxxxxxxxxx> · Fri, 07 Oct 2011 11:55:21 +1300



    On 07/10/11 10:56, Henry Drexler wrote:
    

      On Thu, Oct 6, 2011 at 4:37 PM, Gavin
        Flower <GavinFlower@xxxxxxxxxxxxxxxxx>
        wrote:

        
            On 07/10/11 01:40, Henry Drexler wrote:

              
                I have a workaround to the error/result, but am
                wondering what the result of ts_rank of '1e-020'
                represents?

                
                Here is the original:

                
                select

                ts_rank(to_tsvector('a_a_do_ug_read_retreqmon_ptam'),to_tsquery('a_a_do_ug_read_retrmso.com_ptam'))

                
          I get essentialy the same result with pg 9.1.1

          
          ----------------

           9.99999968e-21

          (1 row)

          
            gavin=>

          
        I am also on "PostgreSQL 9.1.1"
        

        Thanks for posting the   9.99999968e-21, I did not realize
          it was using notation to represent a number, I thought it was
          an error code.  That is good news - thank you for taking the
          time ans showing me your result - that solves it.
      
    
    You're welcome.

    
    It is amazing how 'trivial' things can have such significant effects
    - I once took a few hours to spot an extra ';' in a short C program
    that wasn't doing what it was meant to do!

    
    In postgresql.conf I have:

    extra_float_digits = 3                  # min -15, max 3

    the default is zero.

    
    This might explain the difference in output.