[postgis-tickets] [PostGIS] #2878: winnie coughing up blood when trying to build cunit for twkb

[PostGIS]

#2878: winnie coughing up blood when trying to build cunit for twkb
----------------------+-----------------------------------------------------
  Reporter:  robe     |       Owner:  nicklas      
      Type:  defect   |      Status:  reopened     
  Priority:  blocker  |   Milestone:  PostGIS 2.2.0
 Component:  postgis  |     Version:  trunk        
Resolution:           |    Keywords:               
----------------------+-----------------------------------------------------

Comment(by robe):

 Did a compare of the 2 on my

 {{{
 PostgreSQL 9.4beta2, compiled by Visual C++ build 1800, 32-bit
 }}}
 Here are my results


 {{{
 -- 2 part means varint.h, varint.c (from r12836)
 -- single means varint.h as single file with funcs (from circa r12875)

 -- 2 part runs:  5231 ms (fresh restart),  5121ms (fresh restart), 5663 ms
 (fresh start),
 --      5616ms (fresh start), 4602 ms (warm), 4881ms (warm), 4634ms
 (warm), 4899ms (warm)
 -- single runs: 4991 ms (fresh start), 4649 ms (fresh start), 5554 ms
 (fresh start),  5523ms (fresh start),
 --      5600 ms (warm), 5507 ms (warm), 4555ms (warm), 4540 ms (warm),
 4524 ms (warm)
 DROP TABLE IF EXISTS c;
 create table c as
 select st_astwkb(geom, 0) from a;

 }}}

 and as you said in both cases the explain analyze is consistently the same


 {{{
 -- single run:  execution time (2589.146 ms, plan 1.571ms) (start),
 execution time (2574.146 ms, plan 0.056ms) (warm),
 --              execution time (2623ms, plan 0.049) (warm), execution time
 (2555.248ms, plan 0.049) (warm),

 -- 2 part: execution time (2580.143ms, plan 1.093ms) (start), execution
 time (2612ms, plan 0.049) (warm), execution time (2591.289ms, plan 0.049)
 (warm)
 explain (analyze)
 select st_astwkb(geom, 0) from a;
 }}}


 So even though the create table does seem on average faster with having a
 single file, there is so much noise (particularly with the single) coming
 from somewhere I really can't tell since some runs are actually slower
 than the 2 file.

 I would say if we have a problem, then as you said it probably affects
 every output function we have.  Might as well just ere on the safe side
 and break varint back up and put in a separate ticket to investigate this
 issue.

-- 
Ticket URL: <http://trac.osgeo.org/postgis/ticket/2878#comment:19>
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.