[KinoSearch] remove Hits->seek() [patch]
Nathan Kurz
nate at verse.com
Fri Jul 13 11:49:17 PDT 2007
On 7/13/07, Marvin Humphrey <marvin at rectangular.com> wrote:
> The important bit is that search-time control flow is now less
> convoluted, which should make the code base both easier to spelunk
> and easier to develop.
My one concern about the way I did it (and this is essentially an
artifact from the previous implementation) is that if a search offset
is used, the Hits still contains all the results but just sets its
internal offset (nee tick) to the correct starting position.
Thus this is potentially a trap, since if someone tries uses a
search() offset and then resets the Hits offset to zero to start over,
they won't be getting the same Hit they got the first time. I wasn't
sure if this merited a documentation note.
The alternative would be to build score_docs taking the search offset
into account, so that the earlier results are not even present. For
large offsets this might be helpful, especially when combined with
SearchClient. But it would change behaviour slightly.
> My only feedback is that this is a well-constructed patch which
> demonstrates your familiarity with the material and upholds the
> necessary internal conventions. Nicely done.
Thanks for the compliment. What should I tackle next?
Nathan Kurz
nate at verse.com
More information about the kinosearch
mailing list