On Tue, May 11, 2004 at 02:42:04AM -0700, Markus Peter wrote:
> --On Montag, 10. Mai 2004 20:55 Uhr -0500 David L Norris
> <firstname.lastname@example.org> wrote:
> > As you say, you can store a flag externally to mark deleted items. The
> > front-end could check if the item still exists before adding it to the
> > results. This complicates the front-end slightly but it works great if
> > well implemented.
> Well, it works well until paging comes into play if you want to have a
> consistent interface.
Depends on how big your total result set is. I have one program that
always reads all the results, does filtering and special sorting, and
then caches the result set for paging.
If your result sets are huge then that doesn't work well. But in that
case, it doesn't work that great reading the properties for each result
to find which might be deleted, either.
Might be better to walk the main word index and flag all
word:meta-name:file-number entries for the given file as deleted so the
results are not included in the initial result set in the first place.
Received on Tue May 11 10:23:15 2004