SELECT c FROM mytable WHERE b IS NOT c
Is this what you mean?
Imanuel
Am 29.11.2012 07:45, schrieb YAN HONG YE:
> b c
> 897313897313
> 854809868763
> 854808854809
> 854980854808
> 854974854980
> 854973854974
&
That's better, but only if you know for sure that the data doesn't
contain '-'.
Otherwise, you run into the same problem with the following values:
'1-','2','3' -> '1','-2','3'
Imanuel
Am 25.11.2
ing the expected
update.
Which means with every false hit it has less data to write and thus is
faster because of the false hits (instead of "even with").
Imanuel
Am 25.11.2012 17:13, schrieb Simon Slavin:
>
> On 25 Nov 2012, at 4:11pm, Imanuel wrote:
>
>> Hi Ke
I have read the wrong line in my inbox while typing my reply.
Hopefully you forgive me ;)
Am 25.11.2012 17:11, schrieb Imanuel:
> Hi Keith
>
>> UPDATE user SET a=1,b=2,c=3 WHERE a||b||c!=1||2||3;
>
> It seems to me that this is not reliable.
> Think the the following text
Hi Keith
> UPDATE user SET a=1,b=2,c=3 WHERE a||b||c!=1||2||3;
It seems to me that this is not reliable.
Think the the following text values:
a='12'
b='34'
c='56'
If you want to update these values to:
a='1'
b='2345'
c='6'
Then your statement would not update because '123456' = '123456'.
Greet
se 3.6.22
is way faster than 3.7.14.1 - so it has to be related to any change that
was made.
I'll make more tests if it is only CREATE INDEX that is slowlier, or
other commands, too.
SELECT statements on indexed columns are very fast with both, but maybe
they are simply too fast to not
Sorry if I'm annoying you, but is there any way to have a fast SQLite
3.7.14.1 via DLL or at least to tell why it's that slow?
Imanuel
Am 18.10.2012 17:49, schrieb Imanuel:
> No, I can't - 26s vs 15s (old vs new).
>
> But when I run the test in my Delphi test applicati
rce/SQLite3.pas
I simply have replaced the dll without changing the linking source code
to test with 3.7.14.1 - I hope that's ok.
Imanuel
Am 18.10.2012 16:49, schrieb Dan Kennedy:
> On 10/18/2012 03:32 PM, Imanuel wrote:
>> Ok, here it is (45mb):
>> http://www.file-upload.n
Ok, here it is (45mb):
http://www.file-upload.net/download-6707980/CREATE_INDEX_test.7z.html
Imanuel
Am 18.10.2012 00:37, schrieb Imanuel:
> No, the performance stays the same.
> I have also tried using a big cache_size, but that didn't change
> anything, too.
>
> Yes, I ca
No, the performance stays the same.
I have also tried using a big cache_size, but that didn't change
anything, too.
Yes, I can share the database - it is currently uploading, I will mail
the link tomorrow.
Imanuel
Am 17.10.2012 22:08, schrieb Dan Kennedy:
> On 10/18/2012 01:32 AM,
olumns are presorted.
When I stumbled upon this for the first time, I experienced values of 18
vs 380 seconds but I can't reproduce them now because I don't have that
version of the database any more.
Is there anything I can change?
Imanuel
___
sqlit
11 matches
Mail list logo
