[sqlite] Bug: Successfully committed transaction rolled back after power failure

Please can I formally propose that, for Linux: 1. A Pragma or other compile-time option is created such that SQLITE_OK is not issued on file writes/modifications/deletes until the hardware indicates that all associated Directory syncs, etc., are done. 2. Since the absence of 1. appears to break

[sqlite] Code Cleanup - Variable Definition - malloc.c - sqlite3DbMallocRawNN

Variable Definitions only supported at start of block before code in older (MSVC) compilers. @@ -22479,15 +22479,15 @@ p = sqlite3Malloc(n); sqlite3MemdebugSetType(p, MEMTYPE_HEAP); return p; } SQLITE_PRIVATE void *sqlite3DbMallocRawNN(sqlite3 *db, u64 n){ + LookasideSlot *pBuf;

[sqlite] json_group_array

I just love this list and the contained discussions. Doffing my cap to Keith and his succinct and very useful summary of the subtle differences regarding aggregate type functions. Especially like the nested stuff, which I must admit had not really occurred to me. Nice stuff Keith and thanks for

[sqlite] Bug: Successfully committed transaction rolled back after power failure

On 2/6/16, Bernard McNeill wrote: > Please can I formally propose that, for Linux: > 1. A Pragma or other compile-time option is created such that SQLITE_OK is > not issued on file writes/modifications/deletes until the hardware > indicates that all associated Directory syncs, etc., are done.

[sqlite] Virtual Table Example?

Hello, Does anyone know where to find an example project that creates a SQLite virtual table module using a web service as the data source? Or if anyone has done this type of work, please contact me directly. Thank you, John Stegeman