> The code compiles and runs for me. I am running the regression > test suite as I type this and there are no problems so far. However, > let me restate that I do not have the ability to test the behavior > of SQLite in the presence of virus scanners and Tortoise SVN, so > I am looking for people who do have that capability to help me > by testing these changes thoroughly. Please let me know if the > changes implemented by checking [3200] fail to fix the random > SQLITE_CANTOPEN problem. > I got the changes and recompiled the sqlite3.dll The problem is not there any more: I tested in 2 different machines (both with WinXP and SVN) and I never got any error, while testing at the same time with the previous version I got the error.
Although I cannot say that i am 100% certain that this solution is total, I can see that the error is not happening any more (and reading the changes I can confirm that the patch approaches the problem properly). I will do more tests later and report back should I find any failure. Thank you Richard for your prompt response to this serious issue! Costas