RE: [Dev-C++] (OT:) SQLExecDirect fails on good statements

[Ed C. <em...@to...>]

That would be my suspicion, but I don't know where the problem might lie...
How can I tell which of the libraries included in the Dev-C++/mingw
distribution are not thread-safe?

-Ed

-----Original Message-----
From: Per Westermark [mailto:pw...@ia...] 
Sent: Saturday, October 25, 2003 8:12 PM
To: Ed Cottrell
Cc: dev...@li...
Subject: Re: [Dev-C++] (OT:) SQLExecDirect fails on good statements


I would suspect either your threaded code, or that you are not linking with
a thread-safe library.

/Per W

On Sat, 25 Oct 2003, Ed Cottrell wrote:

> Hi all,
>
> This is probably going to be too abstract to get an answer, but here 
> goes: in an application I'm working on, lots (usually thousands) of 
> database queries are issued of the form
>
> "SELECT division_id FROM my_db.my_table WHERE division='var1' and 
> division_id = var2"
>
> I'm building these queries using sprintf(), then running them with 
> SQLExecDirect; about 99.98% of the time, this works perfectly. The 
> remaining .02% of queries will fail, randomly. For example, the query 
> "SELECT division_id FROM my_db.my_table WHERE division='FOO' and 
> division_id = 1" will work almost every time, but fail .02% of the 
> time.
>
> These queries are being created and executed by (a user-defined number 
> of) threads, but not asynchronously, and the database server is not 
> having any problems. As I've said, the failures are not on (or 
> immediately) following specific queries, but seem to be randomly 
> distributed throughout the thousands of queries being executing. I put 
> in a little error check, so that if SQLExecDirect does not return 
> SQL_SUCCESS, I get a display of the failed query, and I can verify 
> that the data is not being corrupted, I don't have any buffer 
> overflows, etc. The database server is MySQL-max-nt.
>
> I'm sorry, but I can't post full source code, for various reasons. I'm 
> just wondering if anybody knows why perfectly valid SQL statements 
> would fail, once in every 5,000 tries, when the server itself is not 
> being overwhelmed.
>
> -Ed
>