Re: [sqlite] Is it possible to optimize the readperformanceof a C++ app using sqlite pragma journal_mode = wal

2011-05-28 Thread Michael Stephenson
You're welcome. Let's hold judgment on whether any of them turn out to be helpful :o). Increasing the page size might help, it also gives you four times the cache with the same default setting of 2000 cache pages. I would suggest again that you consider bumping that number of cache pages up if

Re: [sqlite] Is it possible to optimize the readperformanceof a C++ app using sqlite pragma journal_mode = wal

2011-05-28 Thread Frank Chang
Michael Stephenson, We want to thank you again for your excellent suggestions regarding how to improve the speed of our Sqlite WAL reads and our deduper prototype. We looked at the SQlite documentation for increasing the Sqlite page size to 4K and an excerpt of our code is shown below. If we

Re: [sqlite] Unlocking the database

2011-05-28 Thread Jean-Christophe Deschamps
>I understand the need for integrity when locking a database, but in >this case I knew that the problem was caused by a (in all honesty, my) >bug. I tried rebooting the machine and it did not unlock the table. Yeah, sometimes Windows will keep a lock on a file beyond reboot, which is

Re: [sqlite] Unlocking the database

2011-05-28 Thread john darnell
Thanks for the tip. I have done as you suggested. > -Original Message- > From: sqlite-users-boun...@sqlite.org [mailto:sqlite-users-boun...@sqlite.org] > On Behalf Of Danny > Sent: Saturday, May 28, 2011 12:46 PM > To: General Discussion of SQLite Database > Subject: Re: [sqlite]

Re: [sqlite] Unlocking the database

2011-05-28 Thread Roger Andersson
On 05/28/11 07:00 PM, Simon Slavin wrote: > SQLite locking is a function of your OS. It's not static things > something like "Byte 4 of the file is set to 'L'", it's transient > things handled with file handles or low level FS stuff. > So reboot. Or possibly find everything that might have

Re: [sqlite] R*-Tree limit

2011-05-28 Thread Richard Hipp
On Fri, May 27, 2011 at 7:59 AM, Kirell Benzi wrote: > Hello, > > I would like to use the R*-Tree for a big number of columns, let's say 50. > However I read in the doc that it was limited to 11. > The number of columns will always be odd: One rowid and the pairs of

Re: [sqlite] Unlocking the database

2011-05-28 Thread Danny
John, I've had nothing but trouble with the Firefox plugin. Download and install the SQLite Expert Personal 3 GUI (free) and see if that does anything for you, or at least gives you better diagnostics. --- On Sat, 5/28/11, Simon Slavin wrote: > From: Simon Slavin

Re: [sqlite] Unlocking the database

2011-05-28 Thread Simon Slavin
On 28 May 2011, at 5:39pm, john darnell wrote: > After the reboot, I tried opening the database in my program as well as > SQLite Manager (the Firefox plugin). Neither worked, I got the SQLITE_BUSY > return code from the SQLite call (I believe it was sqlite3_prepare_v2), and a > long and

Re: [sqlite] Unlocking the database

2011-05-28 Thread Simon Slavin
On 28 May 2011, at 5:02pm, john darnell wrote: > This time I have (I hope) a much simpler question. While debugging my code, > I managed to lock the database for all time. The only way I was able to > unlock the database was by retrieving a copy of the database I had tucked > away for just

Re: [sqlite] Unlocking the database

2011-05-28 Thread john darnell
I was/am using Windows XP SP3. Sorry, but the company is extremely conservative in upgrade policies. Hmmm After the reboot, I tried opening the database in my program as well as SQLite Manager (the Firefox plugin). Neither worked, I got the SQLITE_BUSY return code from the SQLite call (I

Re: [sqlite] Unlocking the database

2011-05-28 Thread Richard Hipp
On Sat, May 28, 2011 at 12:25 PM, john darnell wrote: > I tried rebooting the machine and it did not unlock the table. > That sounds unlikely. What OS are you using? Are you using a non-standard VFS such as "unix-dotfile"? Or are you using a network filesystem of

Re: [sqlite] Unlocking the database

2011-05-28 Thread john darnell
I understand the need for integrity when locking a database, but in this case I knew that the problem was caused by a (in all honesty, my) bug. I tried rebooting the machine and it did not unlock the table. So I guess that means that the answer is "no." R, John > -Original Message-

Re: [sqlite] Unlocking the database

2011-05-28 Thread Roger Binns
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 05/28/2011 09:02 AM, john darnell wrote: > Is there a function, method or process whereby I can unlock the database when > it gets locked? It is locked by another connection or process and you can't ask them to give up locks using the SQLite API.

[sqlite] Unlocking the database

2011-05-28 Thread john darnell
Hello folks. This time I have (I hope) a much simpler question. While debugging my code, I managed to lock the database for all time. The only way I was able to unlock the database was by retrieving a copy of the database I had tucked away for just such contingencies. I looked in Jay

Re: [sqlite] Is it possible to optimize the readperformanceof a C++ app using sqlite pragma journal_mode = wal

2011-05-28 Thread Simon Slavin
On 28 May 2011, at 4:30pm, Frank Chang wrote: > We were wondering if you could tell us what sqlite C/C++ API function to use > to change the SQLIte page size to 4KB. Thank you for all of your help. Page Size is part of the setup of the database file, so the way you set it is this: Use PRAGMA

Re: [sqlite] Is it possible to optimize the readperformanceof a C++ app using sqlite pragma journal_mode = wal

2011-05-28 Thread Frank Chang
Michael Stephenson, Thank you for all of your excellent ideas on increasing the speed of the deduper and the speed of the WAL reads, We will try these ideas. We were wondering if you could tell us what sqlite C/C++ API function to use to change the SQLIte page size to 4KB. Thank you for all