Re: [sqlite] sqlite3_interrupt() for controlled interruption
On 12/3/2013 9:26 PM, Chen, Mi wrote: I want to use sqlite3_interrupt() to interrupt the statement that is executing sqlite3_step() for the first time in the worker. It is mutex protected so I already make sure that when sqlite3_interrupt() is called, my worker thread is running sqlite3_step()... However, the current sqlite3_interrupt() behavior is to interrupt all the statements, including the ones that I suspended and running half-ways in the background. I could make a separate database connection for each query, but the ability for them to share data (e.g., temp tables) were limited. See if sqlite3_progress_handler helps. A progress handler only cancels one sqlite3_step() call. -- Igor Tandetnik ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] sqlite3_interrupt() for controlled interruption
This is exactly what I need. Thanks very much! Mi Chen -Original Message- From: sqlite-users-boun...@sqlite.org [mailto:sqlite-users-boun...@sqlite.org] On Behalf Of Igor Tandetnik Sent: Tuesday, December 03, 2013 7:46 PM To: sqlite-users@sqlite.org Subject: Re: [sqlite] sqlite3_interrupt() for controlled interruption On 12/3/2013 9:26 PM, Chen, Mi wrote: I want to use sqlite3_interrupt() to interrupt the statement that is executing sqlite3_step() for the first time in the worker. It is mutex protected so I already make sure that when sqlite3_interrupt() is called, my worker thread is running sqlite3_step()... However, the current sqlite3_interrupt() behavior is to interrupt all the statements, including the ones that I suspended and running half-ways in the background. I could make a separate database connection for each query, but the ability for them to share data (e.g., temp tables) were limited. See if sqlite3_progress_handler helps. A progress handler only cancels one sqlite3_step() call. -- Igor Tandetnik ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users ___ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
Re: [sqlite] sqlite3_interrupt from another thread
Roger Binns [EMAIL PROTECTED] wrote: We had an earlier discussion about calling sqlite3_interrupt from another thread, which wasn't possible at the time: It looks like the code is fixed for 3.3.12 (and possibly earlier). I just wanted to double check that it is now officially safe to call from another thread. The main documentation doesn't mention anything either way. Go to http://www.sqlite.org/cvstrac/search and do a search for sqlite3_interrupt in tickets and in check-ins. The result is http://www.sqlite.org/cvstrac/search?s=sqlite3_interruptt=1c=1 From this we see that that sqlite3_interrupt() can be called from a separate thread as of check-in [3336] on 2006-07-26 associated with ticket #1897. That was first delivered in version 3.3.7. -- D. Richard Hipp [EMAIL PROTECTED] - To unsubscribe, send email to [EMAIL PROTECTED] -
Re: [sqlite] sqlite3_interrupt()
Rob Richardson [EMAIL PROTECTED] wrote: I have a thread that executes a query that takes about 30 seconds on my test setup. In the field, the query could take several minutes. The user needs to be able to stop this query if it was started by accident. I have a pointer to the sqlite3 object that is running the query inside my thread. In the main GUI thread of the application, I call sqlite3_interrupt(), passing it the stored sqlite3 pointer. But after I do that, the query still stops executing 30 seconds after it started. The interruption seems to have no effect. Is this expected? How soon after I issue sqlite3_interrupt() should a long query stop executing? Or am I doing something wrong? The query should stop immediately. Are you using the latest version of SQLite that supports calling sqlite3_interrupt() from a separate thread, or an older version that might have problems? -- D. Richard Hipp [EMAIL PROTECTED] - To unsubscribe, send email to [EMAIL PROTECTED] -
RE: [sqlite] sqlite3_interrupt()
Dr. Hipp, Thanks for your reply. The question of which version I'm using is up in the air right now. I threw a call to sqlite3_libversion() into my application, and it returned 3.2.1. I asked our lead developer (who's in Indiana while the rest of us are just west of Cleveland, OH) what version we should have, and he said 3.3.5. He sent the source code. I opened my existing copy of sqlite3.h and saw that SQLITE_VERSION was set to 3.3.5. Yet, when I my application, I get 3.2.1. I deleted all copies of sqlite3.lib and sqlite3.dll from my machine, rebuilt sqlite3 from Visual Studio, and still I get 3.2.1. I have no idea where it's coming from, but it seems to be telling me I'm using an old version of SQLite. RobR - To unsubscribe, send email to [EMAIL PROTECTED] -
Re: [sqlite] sqlite3_interrupt()
Rob Richardson [EMAIL PROTECTED] wrote: Dr. Hipp, Thank you very much for your earlier assistance. As you suggested, I downloaded the version 3.3.7 source code and built it into a static library. My application now correctly reports that it is using version 3.3.7. But sqlite3_interrupt() still seems to be doing nothing. I added TRACE statements to my code to show the times things happened, and I included the value of the pointer to the sqlite3 object that was being used when the query starts and as the argument to sqlite3_interrupt(): Can you run your program in a debugger? The sqlite3_interrupt() routine should set the sqlite3.u1.isInterrupted flag. That flag is then tested at various points in the file vdbe.c (whereever you see the macro CHECK_FOR_INTERRUPT) and if it is true, the execution of the statement should stop. The flag is reset once the statement completes. Run your program in a debugger and try to figure out what is going wrong. Let me know if you find a bug in SQLite. (It works when I test it) -- D. Richard Hipp [EMAIL PROTECTED] - To unsubscribe, send email to [EMAIL PROTECTED] -
RE: [sqlite] sqlite3_interrupt()
Dr. Hipp,
Thank you very much for your earlier assistance. As you suggested, I
downloaded the version 3.3.7 source code and built it into a static
library. My application now correctly reports that it is using version
3.3.7.
But sqlite3_interrupt() still seems to be doing nothing. I added TRACE
statements to my code to show the times things happened, and I included
the value of the pointer to the sqlite3 object that was being used when
the query starts and as the argument to sqlite3_interrupt():
GetTrendData() started at Thu Aug 24 16:03:54 2006
; sqlite object is at 056B89E0.
select tag_key,value,datetime(value_timestamp, 'localtime') from
trend_view where trend_key=1 and value_timestamp =
julianday('2006-08-22 16:01:41.000', 'utc') order by value_timestamp
desc
Interrupt requested at Thu Aug 24 16:03:57 2006
for object at 056B89E0.
GetTrendData() finished at Thu Aug 24 16:05:07 2006
Rob Richardson
RAD-CON INC.
-
To unsubscribe, send email to [EMAIL PROTECTED]
-
RE: [sqlite] sqlite3_interrupt()
I think I've found the problem.
The next step in debugging was to find the actual sqlite3 library calls
I was making, just to make sure the delay wasn't somewhere else in my
code. Here's the trace statements:
GetTrendData() started at Thu Aug 24 16; sqlite object is at 056B9570.
select tag_key,value,datetime(value_timestamp, 'localtime') from
trend_view where trend_key=1 and value_timestamp =
julianday('2006-08-22 16C
SqlOpen(): First sqlite3_step() called at Thu Aug 24 16:30:52 2006
; sqlite object is at 056C20E0.
Interrupt requested at Thu Aug 24 16:30:54 2006
for object at 056B9570.
CSqlOpen(): Back from first sqlite3_step() call at Thu Aug 24 16:32:04
2006
; sqlite object is at 056C20E0.
GetTrendData() finished at Thu Aug 24 16
The object pointer reported in the second statement is not the same as
the one in the first statement. The second statement's pointer is the
one that was used in the call to sqlite3_prepare(). The
sqlite3_interrupt() pointer had better be the same as the one in
sqlite3_prepare(), and it wasn't.
Bottom line: I was trying to interrupt the wrong sqlite3 object.
Once I figure out how to get the correct sqlite3 object to use in the
interrupt call, I'll let you know how it works.
Rob Richardson
RAD-CON INC.
-
To unsubscribe, send email to [EMAIL PROTECTED]
-
Re: [sqlite] sqlite3_interrupt() and threads
Am 26.07.2006 um 01:06 schrieb [EMAIL PROTECTED]: Michael Scharf [EMAIL PROTECTED] wrote: Rob, I notice in the documentation that the sqlite3_progress_handler() method is marked experimental. Is that significant? No idea, that's a question Richard Hipp may answer.. I need to remove the experimental marking. Perhaps somebody could create a ticket to remind me ;-) While you are at it, consider creating an enhancement ticket to allow sqlite3_interrupt to be called from a different thread. Tickets #1897 and #1898. I saw that you already committed code to make sqlite3_interrupt thread-safe, but created a ticket anyway to make sure it's tracked. /jum
Re: [sqlite] sqlite3_interrupt() and threads
Rob Richardson wrote: [...] Can sqlite3_interrupt() be called from a different thread than the sqlite3_step() that I want to interrupt, or is there some other mechanism I can use, or is there no way to do this? This came up a couple of weeks ago. I don't have the message here to quote but you can see it at: http://www.mail-archive.com/sqlite-users@sqlite.org/msg16361.html Martin
RE: [sqlite] sqlite3_interrupt() and threads
Thank you for pointing me to that article. That leads to two more questions: What alternative do I have? If I abort the thread that is running the query, will the query stop? And, if sqlite3_interrupt() has to be issued from the same thread that is using the database, what is it designed to be used for? OK, three questions: Is there a way to run sqlite3 queries asynchronously? Thanks again! Rob Richardson Rad-Con, Inc.
Re: [sqlite] sqlite3_interrupt() and threads
Hi Richard, I use a progres_handler: http://www.sqlite.org/capi3ref.html#sqlite3_progress_handler I set it up to be called every 1 ticks (2nd argument) or so and I use and the void* to points to a data structure that contains a boolean isCanceled. If another thread wants to cancel the worker thread it sets the isCanceled to true. The progress callback checks the isCanceled and returns 0 if set. Then the query gets aborted... (I'm using this in java and it works just fine) Michael Thank you for pointing me to that article. That leads to two more questions: What alternative do I have? If I abort the thread that is running the query, will the query stop? And, if sqlite3_interrupt() has to be issued from the same thread that is using the database, what is it designed to be used for? OK, three questions: Is there a way to run sqlite3 queries asynchronously? Thanks again! Rob Richardson Rad-Con, Inc. -- http://MichaelScharf.blogspot.com/
RE: [sqlite] sqlite3_interrupt() and threads
Great! That looks like exactly what I need. Thanks very much! Rob Richardson RAD-CON INC. -Original Message- From: Michael Scharf [mailto:[EMAIL PROTECTED] Sent: Tuesday, July 25, 2006 10:45 AM To: sqlite-users@sqlite.org Subject: Re: [sqlite] sqlite3_interrupt() and threads Hi Richard, I use a progres_handler: http://www.sqlite.org/capi3ref.html#sqlite3_progress_handler I set it up to be called every 1 ticks (2nd argument) or so and I use and the void* to points to a data structure that contains a boolean isCanceled. If another thread wants to cancel the worker thread it sets the isCanceled to true. The progress callback checks the isCanceled and returns 0 if set. Then the query gets aborted... (I'm using this in java and it works just fine) Michael Thank you for pointing me to that article. That leads to two more questions: What alternative do I have? If I abort the thread that is running the query, will the query stop? And, if sqlite3_interrupt() has to be issued from the same thread that is using the database, what is it designed to be used for? OK, three questions: Is there a way to run sqlite3 queries asynchronously? Thanks again! Rob Richardson Rad-Con, Inc. -- http://MichaelScharf.blogspot.com/
RE: [sqlite] sqlite3_interrupt() and threads
Michael, I notice in the documentation that the sqlite3_progress_handler() method is marked experimental. Is that significant? Rob
Re: [sqlite] sqlite3_interrupt() and threads
Rob, I notice in the documentation that the sqlite3_progress_handler() method is marked experimental. Is that significant? No idea, that's a question Richard Hipp may answer.. Michael -- http://MichaelScharf.blogspot.com/
Re: [sqlite] sqlite3_interrupt() and threads
Michael Scharf [EMAIL PROTECTED] wrote: Rob, I notice in the documentation that the sqlite3_progress_handler() method is marked experimental. Is that significant? No idea, that's a question Richard Hipp may answer.. I need to remove the experimental marking. Perhaps somebody could create a ticket to remind me ;-) While you are at it, consider creating an enhancement ticket to allow sqlite3_interrupt to be called from a different thread. -- D. Richard Hipp [EMAIL PROTECTED]
[sqlite] unsubscribe me please RE: [sqlite] sqlite3_interrupt() and threads
Chunde Shi [EMAIL PROTECTED] 703 882 1466 703 882 2325 From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Sent: Tue 7/25/2006 7:06 PM To: sqlite-users@sqlite.org Subject: Re: [sqlite] sqlite3_interrupt() and threads Michael Scharf [EMAIL PROTECTED] wrote: Rob, I notice in the documentation that the sqlite3_progress_handler() method is marked experimental. Is that significant? No idea, that's a question Richard Hipp may answer.. I need to remove the experimental marking. Perhaps somebody could create a ticket to remind me ;-) While you are at it, consider creating an enhancement ticket to allow sqlite3_interrupt to be called from a different thread. -- D. Richard Hipp [EMAIL PROTECTED]
