[ZODB-Dev] On-disk index for FileStorage

[Jeremy Hylton]

>>>>> "TD" == Toby Dickenson <tdickenson@geminidataloggers.com> writes:

  TD> On Tuesday 10 December 2002 9:08 pm, Greg Ward wrote:
  >> Many months ago, someone posted a patch that fixed FileStorage so
  >> it no longer read in its entire index at open(), or wrote it out
  >> in one big spew at close(), or kept the whole thing in memory all
  >> the time.
  >>
  >> Who was this?  Is this patch still alive?

  TD> are you thinking of jims 'fsindex' patch? it reads the whole
  TD> index into memory, but uses a nested btree instead of dict to
  TD> reduce memory linearly.  this is standard in FileStorage now.

I don't know if anyone noticed the checkin, but it only became
standard last week!  The fsIndex code was developed on a branch of a
branch, and when it was merged all the way to the trunk the fsIndex
change was lost.

Jeremy