--- !ditz.rubyforge.org,2008-03-06/issue 
title: free internal enumerator entries as soon as they've been enumerated
desc: |-
  Currently the internal enumerator holds the entire enumaration sequence
  until the internal enumerator is freed.  Because there is no way to
  enumerate a particular sequence twice, the enumerated entries could be
  overwritten with e.g. "undefined" values as the enumeration proceeds,
  generating string free operations steadily instead of as on massive
  chunk as the enumeration finishes.
type: :task
component: duk
release: 
reporter: sva <sami.vaarala@iki.fi>
status: :unstarted
disposition: 
creation_time: 2013-12-17 11:18:51.672163 Z
references: []

id: 149481dcf4856c1108d2ef4e17feb2b448fa3dbe
log_events: 
- - 2013-12-17 11:18:51.864087 Z
  - sva <sami.vaarala@iki.fi>
  - created
  - ""
- - 2013-12-17 11:19:12.792144 Z
  - sva <sami.vaarala@iki.fi>
  - commented
  - Update hobject-design.txt too.
- - 2013-12-17 12:45:28.370024 Z
  - sva <sami.vaarala@iki.fi>
  - commented
  - |-
    Actually enumeration goes through the *keys* (not values) of the internal
    enumerator.  The keys cannot be replaced with "undefined".  They could be
    replaced with NULL, but this would lead to the risk of the keys being
    compacted and the "_next" internal property being incorrect.
    
    A practical approach would be to replace each key with the empty string as
    soon as it was enumerated.  This would create duplicate keys in the entry
    part, however, something that is not normally allowed so it would need
    careful consideration.
- - 2013-12-19 00:58:14.586414 Z
  - sva <sami.vaarala@iki.fi>
  - assigned to release v0.10 from v0.9
  - ""
- - 2014-02-18 13:22:36.989355 Z
  - sva <sami.vaarala@iki.fi>
  - unassigned from release v0.10
  - ""