On 31 May 2009, at 5:53pm, Jim Wilcoxson wrote: > the real point here is that Python and SQLite aren't doing real > transactions
But they /are/ real transactions. You write rows, and read them back, and the values are there. Use SQL commands in the way they're meant to be used and the DBMS does what you expect it to do. It doesn't crash your application, and it doesn't return the wrong values. The fact that instead of being written as magnetic domains on a disk surface your data is floating around several layers of caching may be a problem if your computer crashes, but only then. It's not as if you have several computers accessing the hard disk and they were each getting different answers. > But by disabling the IDE hard drive write cache: I see what you're saying, but the IDE hard drive is unmistakably part of the storage system. It may not be the write-head and platters but it's part of the hard drive system. My feeling is that even if you know from spin speed that it's impossible to write that fast, you can't really start splitting the mass storage system up in that way, you just have to accept that a 'write' command does writing. The subject header says 'Slow Transaction Speed'. You get fast speed when you let all the caches do what they normally do (default settings). Compare those with other implementations of SQL and SQLite does pretty well. I don't see any reason to complain. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users