On Mon, Jun 24, 2013 at 10:28 AM, Andrew Beal <ab...@whoi.edu> wrote:
> Hello All, > > I am trying to implement SQLite on our embedded platform. Since I am in > the process of implementing the VFS, I was trying to give our other > developers the chance to use the in-memory database while we implement the > VFS, however I have noticed that the open code for a database refers > finding the default VFS which then fails since no VFS is present. > > I suspect that this is a bug since it prevents the use of the in-memory > database while the VFS isn't implemented or present. > The VFS is used for things other than file I/O. Examples: the xSleep() method used to implement sqlite3_sleep() and the xCurrentTimeInt64() method used by the date and time functions to get the current system time and the xRandomness() method that is used to seed SQLite's internal PRNG. So a VFS is necessary even if the ":memory:" database is being used. If os_unix.c won't even link on your system, then I suggest you create a quick VFS that has stub functions for all of the disk I/O routines - functions that always return SQLITE_ERROR. You can probably use one of the test VFSes like test_vfstrace.c or test_onefile.c as a template. Make a copy of one of those files, then go into the body of each disk I/O method, delete the existing code, and replace the deleted code with "return SQLITE_ERROR;". -- D. Richard Hipp d...@sqlite.org _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
