A full table scan of 16 million rows and 55 columns on Vista with
FILE_FLAG_RANDOM_ACCESS finished in 600 seconds (cold) and 499 seconds
(warm) and turned Vista into a brick. The same test with the flag removed
ran in 119 seconds (cold) and 99 seconds (warm), and caused no adverse
problems with
On Wed, Sep 17, 2008 at 06:00:45PM -0700, Roger Binns scratched on the wall:
> The second is that SQLite when opening a file under Windows explicitly
> tells Windows that the file will be used for random access even though
> that is not the case. Windows uses this hint to override its builtin
>
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Brown, Daniel wrote:
> Has anyone tried to replicate this bug on WindowsXP too?
There are two separate bugs here. One is that when Vista is told a file
is used for random access and then there is a lot of file access, it may
use a little too much
Has anyone tried to replicate this bug on WindowsXP too?
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Roger Binns
Sent: Wednesday, September 17, 2008 4:24 PM
To: General Discussion of SQLite Database
Subject: Re: [sqlite] Vista frustrations
-BEGIN
I lost my note on what I did to make this work. I've rearranged my
computers for work and installed sqlite3 on Ubuntu. When I installed
sqlite3, I did ./configure --disable-shared. When I recompile my working
code on the new machine:
gcc -static -L/usr/local/lib -I/usr/local/include mycode.o
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Robert Simpson wrote:
> Ideally, at least on non-CE platforms, I'd like see SQLite not give the OS
> any hints about caching. However, I'm not sure what kind of performance hit
> (if any) that would have on Windows. It's already been proven that
>
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Fred Williams wrote:
> Have you ever actually used a version of Windows?
Windows 1.0 (once), Windows 2/286, Windows 3, 3.1, 3.11, 95, 98, ME,
2000, XP, 2003 and Vista.
> ANY OS that attempts to read in a xGigibyte file into real memory to the
>
Thinking through how to protect integrity of rows in an "optimistic
locking" scenario, I can't see a way around either
- comparing all original data values on posting an updated record to see
if the underlying record had been altered after being read
- comparing original record generation
I am running vista my personal pc. I'll try to take a look into how
not having this flag affects performance when I get home.
On Wed, Sep 17, 2008 at 5:29 PM, Robert Simpson <[EMAIL PROTECTED]> wrote:
> Ideally, at least on non-CE platforms, I'd like see SQLite not give the OS
> any hints about
Ideally, at least on non-CE platforms, I'd like see SQLite not give the OS
any hints about caching. However, I'm not sure what kind of performance hit
(if any) that would have on Windows. It's already been proven that
providing the hint on WinCE is beneficial.
-Original Message-
From:
Have you ever actually used a version of Windows?
ANY OS that attempts to read in a xGigibyte file into real memory to the
detriment of the entire system load is not working correctly. Call it a bug
or a feature it still sucks.
I expect nothing less from Microsoft with each new version. I
can multiple page cache be configured ?
For instance say I have one thread that uses db connections that have 2k pages
and a different thread/connection that uses 4k pages. Can independent page
caches be created for each?
Also is there any way to have Sqlite report back the pagecache size for
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Robert Simpson wrote:
> The purpose of a cache is to improve performance and responsiveness. Any
> cache that uses all physical memory, forces all other apps to the paging
> file
All current operating systems do this, using heuristics to determine
Hi list,
I'm trying to get sqlite to utilize the pagecache and scratch buffers.
But when the code exits it returns pagecache overflow and scratch buffer
overflow .
The to configure the pagecache and scratch are done in the main section of
code as follows:
Then connections to sqlite are made
Sure. I just tell to do this test to check if the bug is related to this
component, since it debuted on Vista.
Virgilio Alexandre Fornazin
High performance and realtime systems development
Rua Brigadeiro Vicente Faria Lima, 268
Bela VistaLeme-SPCEP 13611-485
Phone: +55 19 3571-5573
2008/9/17 Dan <[EMAIL PROTECTED]>:
>
>
>>
>> On 3.4.2 we get:
>> 3.0|3660.5|3
>> 6.0|1360.3|6
>>
>> On 3.6.2 we get:
>> |5020.8|3
>>
>
> 3.6.2 has a bug involving DISTINCT or GROUP BY queries that use
> expression aliases (AS clauses) in the select-list. Problem is fixed
> in cvs:
>
>
> Hi All,
> We have been using SQLite 3.4.2 for some time. On investigating
> upgrading to 3.6.2, we found that different results were produced for
> one query.
> The following illustrates:
>
...
>
> On 3.4.2 we get:
> 3.0|3660.5|3
> 6.0|1360.3|6
>
> On 3.6.2 we get:
> |5020.8|3
>
3.6.2 has a
Scott,
thanks again for the many lines of good hints and help.
I think I will simply ignore the problem until I have real end user
experience and see where it will or could end. The FullTextModul more or
less has no whatever specific database interface, it is simply an abstract
black hole that
Hi All,
We have been using SQLite 3.4.2 for some time. On investigating
upgrading to 3.6.2, we found that different results were produced for
one query.
The following illustrates:
CREATE TABLE tst1( tst1Id INTEGER, width REAL, thickness REAL );
CREATE TABLE tst2( tst2Id INTEGER,
Hi All,
I have c++ application which is used SQLite 3.5.9.
Occasionally the application is crash on
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
sqlite3_finalize(pStmt);
Is
SQLite in general and the .Net provider in particular are most often
shipped as components of other applications. I dont think having
developers tell their end users to disable superfetch is a viable
solution. As much as I hate to propose this maybe a runtime check is
in order to see what the OS
Note that Windows Server 2008 use the same 'core' as Windows Vista.
If you´re detecting and redirecting by using GetVersion() or other
approach you might test for Server 2008 too.
Virgilio Alexandre Fornazin
High performance and realtime systems development
Rua Brigadeiro Vicente Faria Lima,
I've run the tests with superfetch and prefetch disabled and enabled.
Results are consistent with or without these running. The only thing that
has any affect is the FILE_FLAG_RANDOM_ACCESS flag. And only on Vista.
For now I'm thinking of overriding the default Windows VFS and redirecting
the
Could not this bug be related with Vista feature called 'Superfetch' ?
It tries to keep in memory the most accessed files for user, avoiding
disk for read access.
If you disable (or stop) this service, the problem remains or not ?
Virgilio Alexandre Fornazin
High performance and realtime
On Wed, Sep 17, 2008 at 01:17:51AM -0700, Roger Binns scratched on the wall:
> Robert Simpson wrote:
> > To me this seems like an obvious bug in Vista,
>
> Actually I'd argue that it is behaving as designed.
You could argue it is behaving as designed, but I'd still argue it is
behaving
Possibly the reason a large number of us are still running Win 2000 :-)
It seems to be the least Windows like Windoze ever released...
When follow on support degrades to an untenable level, I'll either switch to
Linux with a Windoze emulator or maybe run whatever the future ruler of the
Universe
It's been reproduced on Vista x64 and x86 OS's. I don't have a pre-SP1
installation to verify if it happens against pre-patched versions.
Getting a database should be easy enough ...
CREATE TABLE FOO(a, b, c, d, e, f);
Then do this about 16 million times or so ... whatever it takes to pump the
The real frustration is that this seems to be a rather obvious bug in Vista,
and definitely not SQLite's responsibility. IMO setting the flag is the
"right thing to do" -- but at the same time, I don't expect any favors from
Microsoft in fixing this any time soon. Meanwhile all those poor Vista
On Wed, Sep 17, 2008 at 12:15 AM, Holger Lembke <[EMAIL PROTECTED]> wrote:
>>> So it reads: grows from 3 MB to 85 MB.
>> That IS an important distinction!
>
> So what to do?
>
> a.) let it grow.
> b.) do optimize() (but how often and how to detect when it is needed?)
> c.) create a new database by
The purpose of a cache is to improve performance and responsiveness. Any
cache that uses all physical memory, forces all other apps to the paging
file and turns the operating system into a brick is definitely not working
as designed.
-Original Message-
From: [EMAIL PROTECTED]
On Wed, Sep 17, 2008 at 11:44:38AM +0530, Aravinda babu wrote:
> Is there any easy way to get the last row in the table ?
what do you mean by last? last by row id?
then
select * from t1 where oid = (select max(oid) from t1);
should work
___
2008/9/17 Bjorn Rauch <[EMAIL PROTECTED]>:
>> From: Simon Davies <[EMAIL PROTECTED]>> Date: 2008/9/16> Subject: Re:
>> [sqlite-announce] Foreign key .
.
.
.
> Yes, indeed it works this way. But I am executing these commands from a .NET
> application using SQLite.NET (SQLite for ADO.NET 2.0,
>
> From: Simon Davies <[EMAIL PROTECTED]>> Date: 2008/9/16> Subject: Re:
> [sqlite-announce] Foreign key trigger in transaction> triggers wrongly> To:
> [EMAIL PROTECTED], [EMAIL PROTECTED]> > > 2008/9/16 Björn Rauch <[EMAIL
> PROTECTED]>:> > Hello,> .> .> .> > This looks to me like the trigger
"(*) Sun's ZFS can even detect strided sequential access - ie reading X
amount of data every Y kilobytes."
... and so can the NT cache manager since the very first Windows NT
release ;-) It's good to see that people are finally adapting these
features 15 years later.
F.
On Wed, Sep 17, 2008 at
I C sharp code under the silver moonlight.
On Wed, Sep 17, 2008 at 1:21 AM, Andy Allord <
[EMAIL PROTECTED]> wrote:
> How about Flex, AIR and a P backend!
>
>
> On Sep 16, 2008, at 4:33 PM, Javier Julio wrote:
>
> > Um... excuse me! You can all move to the side and make room for Flex
> > and
The Tcl interface is, as others have mentioned, the original one and suported
by the sqlite developers. it is:
1) very easy to use
2) up to date
3) more complete than alternatives
4) flexible
and given Richard Hipp's comment on the amount of Tcl code used to test
sqlite, that looks very
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
Robert Simpson wrote:
> To me this seems like an obvious bug in Vista,
Actually I'd argue that it is behaving as designed. Generally
filesystem code will try to detect what is going on under the hood. In
particular if it looks like you are doing
> I ran some tests using the command-line sqlite3.exe, and observed that
> Windows Vista (SP1) is actually trying to cache the entire 5gb file
> into
> memory during the table scan! The system slows to a complete crawl and
> becomes unresponsive. The sqlite3.exe's memory remains very minimal,
>
Scott,
>> So it reads: grows from 3 MB to 85 MB.
> That IS an important distinction!
So what to do?
a.) let it grow.
b.) do optimize() (but how often and how to detect when it is needed?)
c.) create a new database by reindexing my data. (same problems as b.)
d.) ignore the problem until
Am 16.09.2008 um 16:51 schrieb Dan:
>
> On Sep 16, 2008, at 4:44 PM, Lothar Behrens wrote:
>
>> Hi,
>>
>> I do have any unfinalized statements in my application when compiled
>> on Windows, but not on Mac OS X.
>>
>> Is there any difference I am missing to attent for ?
>>
>> How could I see,
Yes,
that's what I expected, since my debugging attempts into my sqlite
library version always failed.
And it makes it clear why there are these differences.
Thanks
Lothar
Am 16.09.2008 um 17:00 schrieb Jay A. Kreibich:
> On Tue, Sep 16, 2008 at 11:44:24AM +0200, Lothar Behrens scratched
Aravinda babu wrote:
> Is there any easy way to get the last row in the table ?
What do you mean by "last row"?
Do you mean you want to get back the row that was most recently inserted?
If so, then the table needs to have a column containing info about
insertion order of rows, such as a serial
select * from t order by rowid desc limit 1;
Thanks,
Trey Mack
-Original Message-
From: [EMAIL PROTECTED]
[mailto:[EMAIL PROTECTED] On Behalf Of Aravinda babu
Sent: Wednesday, September 17, 2008 2:15 AM
To: sqlite-users@sqlite.org
Subject: [sqlite] Getting the last row
Hi all,
Is there
Hi all,
Is there any easy way to get the last row in the table ?
Thanks,
Aravind.
___
sqlite-users mailing list
sqlite-users@sqlite.org
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
44 matches
Mail list logo
