Thanks Kees,
In fact using integer instead of string gives very similar results.
Dropping the unique index constraint and then creating it again when needed
sounds interesting, as insertion without the unique index is satisfactory and
constact over time.
I tried this, but got a trivial
The query language that fts understands doesn't need to be the same as the
query language the user types. In fact, it is often advantageous to have the
user visible query language be much more lenient than what fts wants to
consume. Most sqlite apps don't force users to type sql!
Linus
On Fri, 14 Sep 2007 23:20:53 +0300, you wrote:
>
>Dear All,
>
>I have been struggling with the performance of insertion in sqlite.
>
>Here we have a very simple case :
>
>A table with an integer autoincrement primary key and a text field that is
>unique.
>
>CREATE TABLE my (id PRIMARY KEY,
On 9/14/07, Asif Lodhi <[EMAIL PROTECTED]> wrote:
> Hi,
>
> I've been just glancing over the Unicode related posts. However, I
> would like to know how it's possible for me to insert text in
> different languages in an Sqlite database. Can you fixate the "type"
> of the language for any particular
Hi,
I've been just glancing over the Unicode related posts. However, I
would like to know how it's possible for me to insert text in
different languages in an Sqlite database. Can you fixate the "type"
of the language for any particular database? I would be thankful if
any of you guys could
"Samuel R. Neff" <[EMAIL PROTECTED]> wrote:
> The /10 syntax makes sense to programmers but I think users are going to
> forget it pretty quickly. Same with "OR" an "NEAR" being required to be all
> caps (I didn't know that). Ideally the UI an application exposes would show
> the user that OR
I tried eliminating sqliteInt.h and replacing with explicit declarations
of i64 and u8. That part worked fine, but func_ext.c also uses
sqliteMalloc
which is also defined in sqliteInt.h which led me down a rabbit hole of
pulling
more and more from sqliteInt.h, and I still can't eliminate the
The /10 syntax makes sense to programmers but I think users are going to
forget it pretty quickly. Same with "OR" an "NEAR" being required to be all
caps (I didn't know that). Ideally the UI an application exposes would show
the user that OR and NEAR were interpreted as keywords and not tokens
Dear All,
I have been struggling with the performance of insertion in sqlite.
Here we have a very simple case :
A table with an integer autoincrement primary key and a text field that is
unique.
CREATE TABLE my (id PRIMARY KEY, url);
CREATE UNIQUE INDEX myurl ON my(url);
My application
"Scott Hess" <[EMAIL PROTECTED]> wrote:
>
> One thing I'll think on in the background as a how-to-integrate
> question is the balance between sophistication for query experts
> versus the approachability for non-experts. For some systems, having
> things like proximity queries complicates the
[Just some background, while I'm thinking about it.]
Google has placeholder syntax, if you search for 'full * search', the
first hit is a Wikipedia article on 'full text search'. It's not the
same, but it is in a similar ballpark. For instance, you might have
'full */10 search' (Google doesn't,
Thanks for the quick response, much appreciated. Guess I better go and look
at the query parser.
Thanks again
Mike
-Original Message-
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]
Sent: 14 September 2007 19:22
To: sqlite-users@sqlite.org
Subject: Re: [sqlite] Adding additional
"Mike Marshall" <[EMAIL PROTECTED]> wrote:
>
> 1) We need to be able to index items such as AT, this seems like
> it's a case of replacing the default tokeniser with our own implementation
Correct.
>
> 2) A NEAR query operator so we can do things like 'foo NEAR10 bar'
> which will
Hi all
Currently we use SQLite in several of our products (feel free to add us to
the companies page, Lexalytics - www.lexalytics.com) and are now looking at
replacing the fulltext engine in one of those projects with FTS. We've done
some testing etc (on fts2) and are pretty sure that it can
Has anyone successfully got either of these to build on Solaris 10,
using the gcc that ships with it? I've tried on 10u4 on x86 and
(after fixing the known problem with B_FALSE/B_TRUE for 3.4.2) they
both failed sometime while linking. I just did a
./configure --prefix/what/ever
with no
15 matches
Mail list logo
