[postgis-devel] [PostGIS] #703: measures test failure

[PostGIS]

#703: measures test failure
---------------------+------------------------------------------------------
 Reporter:  strk     |       Owner:  pramsey      
     Type:  defect   |      Status:  new          
 Priority:  medium   |   Milestone:  PostGIS 2.0.0
Component:  postgis  |     Version:  trunk        
 Keywords:           |  
---------------------+------------------------------------------------------

Comment(by nicklas):

 Yes, nasty I confirm.

 I tried with intermediate variable and it doesn't help. How come?
 I don't understand it.

 The row that makes difference between the systems is:

 theP.x = A->x+r*(B->x-A->x);

 A->x holds the value -40.0 and r*(B->x-A->x) is supposed to be 40.0

 So I putted r*(B->x-A->x) in variable t so I then tried
 theP.x = A->x+t;

 that gives the same error. So I substituted A->x with -40.0 =>
 same error.

 So I substitute t with 40.0 and the error disappears.

 So t holds a value not equal to 40.0 but if I inspect the value like
 lwnotice("t=%.17e",t);
 I get just 4.00000000000000000+01

 Another interesting thing is that using an intermediate variable removes
 the error in the stand alone version like:

 {{{

 #include <stdio.h>

 main()
 {
 double r=1100.0/1375.0;
 double t=r*(50);
 double px = -40.0+r;
 printf("r=%.16e\npx=%.16e\n",r, px);
 }

 }}}

 So outside postgresql/postgis it works to put the value in an intermediate
 variable to trunkate it. But not inside Postgis.
 Why?
 How come?
 And where is the extra bytes saved inside postgis, when I have saved it to
 a double?

 Any hint is appreciated.

 Thanks
 Nicklas

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/703#comment:15>
PostGIS <http://trac.osgeo.org/postgis/>
The PostGIS Trac is used for bug, enhancement & task tracking, a user and developer wiki, and a view into the subversion code repository of PostGIS project.