[Lifelines-dev] suggested enhancements

[prapp <pr...@er...>]

Some enhancement suggestions, for whatever that is worth.

1) Modifying name & refn (especially) refn records to be
internally sorted, so as to allow binary search on them - 
for scalability to large datasets. (I think they use internal
linear traversal right now - but won't swear to it.)


2) Adding double-buffering to protect db against 
corruption if the program is terminated in the middle
of an update of some kind. Double-buffering the record
to update is pretty easy - double-buffering the tree nodes
as they are being updated is a bit more annoying, think I.
I assume that a fixed db file should be used for double-
buffering to increase its chance of remaining in the disk
cache, as it will surely be accessed all the time...

(Of course I include modifying the open db code to
check for outstanding operations)