[GHC] #4225: RTS -i (heap profiling interval) documentation could be clearer (was: RTS option to remove profiling overhead from heap profile time stamps)

[GHC]

#4225: RTS -i (heap profiling interval) documentation could be clearer
---------------------------------+------------------------------------------
    Reporter:  ClaudiusMaximus   |       Owner:                   
        Type:  bug               |      Status:  new              
    Priority:  normal            |   Component:  Profiling        
     Version:  6.12.3            |    Keywords:                   
          Os:  Unknown/Multiple  |    Testcase:                   
Architecture:  Unknown/Multiple  |     Failure:  Documentation bug
---------------------------------+------------------------------------------
Changes (by ClaudiusMaximus):

  * failure:  None/Unknown => Documentation bug
  * type:  feature request => bug


Comment:

 Ok, I've done some brief testing and it seems that not specifying -i at
 all does give much saner behaviour of census at naturally-occuring GC
 (though, -s reports around 75 samples and the .hp has fewer, around 57
 samples).

 Perhaps when the documentation is updated to clarify this point there
 should be an additional note warning that specifying -i will force
 additional GC which might affect program behaviour?

 Thanks for aiding my understanding of all this :)

 PS: I found a link to this page but it does not yet contain any
 information:
 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Profiling/Heap
 linked from:
 http://hackage.haskell.org/trac/ghc/wiki/Commentary/Rts/Storage/Slop
 (which implies that census requires a freshly GC'd heap)

-- 
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4225#comment:5>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler