use the results of
all this CPU labor.
-Ursprüngliche Nachricht-
Von: sqlite-users [mailto:sqlite-users-boun...@mailinglists.sqlite.org] Im
Auftrag von Jens Alfke
Gesendet: Samstag, 15. Oktober 2016 20:01
An: SQLite mailing list
Betreff: [sqlite] Any performance penalty for SELECTing more
On 15 Oct 2016, at 7:34pm, Jens Alfke wrote:
> I do, actually, which is why I asked. One of the columns is a blob holding a
> JSON document that can be arbitrarily large. It sounds like including this
> column in the SELECT clause will cause the entire blob to be read from disk?
> I definitel
Jens Alfke wrote:
> In a simple SELECT query of a single table, using the C API,
> is there any difference in performance for requesting more or
> fewer columns of the table in the result? Or is the performance
> penalty only incurred when actually reading the column values?
There can sometimes
Jens Alfke wrote:
> What if I’ve enabled memory-mapping? In that case will the register
> merely point to where the blob data is mapped into memory
This would not work because the data might not be in consecutive pages.
(The database file format was not designed for memory mapping.)
Regards,
Cle
> On Oct 15, 2016, at 11:10 AM, Clemens Ladisch wrote:
>
> In practice, this does not matter unless you have large strings/blobs
> that must be read from overflow pages.
I do, actually, which is why I asked. One of the columns is a blob holding a
JSON document that can be arbitrarily large. It
Jens Alfke wrote:
> is there any difference in performance for requesting more or fewer
> columns of the table in the result? Or is the performance penalty only
> incurred when actually reading the column values?
During the sqlite3_step() call, all values in the SELECT clause are
copied into tempo
In a simple SELECT query of a single table, using the C API, is there any
difference in performance for requesting more or fewer columns of the table in
the result? Or is the performance penalty only incurred when actually reading
the column values?
For example, lets say a table has 26 columns,
7 matches
Mail list logo