[Ocaml-pxp-users] Non-tail-recursive function find_all_elements causes a crash

[Richard Jones]

Just a note here.

I'm using PXP 1.1.95, and I've been tracking a nasty little bug in my
code down much of the afternoon.  It turned out in the end to be due
to a stack overflow in Pxp_document.find_all_elements (actually in
Pxp_document.find_all:search_flat).  My data set contains O(1 million)
records which mostly match the predicate, so quickly causing the
process to die.

My next question will be why, instead of throwing a Stack_overflow
exception, the process goes into an infinite loop throwing SIGSEGV,
but I suspect that's a more fundamental compiler bug and nothing to do
with PXP.

Rich.

let find_all ?(deeply=false) f base =
  let rec search_flat children =
    match children with
        [] -> []
      | n :: children' ->
          if f n then n :: search_flat children' else search_flat children'
  in
  let rec search_deep children =
    match children with
        [] -> []
      | n :: children' ->
          let rest =
            search_deep (n # sub_nodes) @ search_deep children' in
          if f n then
            n :: rest
          else
            rest
  in
  (if deeply then search_deep else search_flat)
  (base # sub_nodes)
;;

let find_all_elements ?deeply eltype base =
  find_all
    ?deeply:deeply
    (fun n ->
       match n # node_type with
           T_element name -> name = eltype
         | _              -> false)
    base
;;


-- 
Richard Jones, CTO Merjis Ltd.
Merjis - web marketing and technology - http://merjis.com
Team Notepad - intranets and extranets for business - http://team-notepad.com