hi,
[repost after earlier posting was mangled up with some HTML tags]
I did some tests with the in-memory database and didn't find any way to use
the in-memory database from multiple db handles in the same process. Having
this capability would enable a multithreaded application to have e.g. multiple
read transactions (each with its own db handle) to run in parallel.
As far as I could see this is not possible. Correct? The implementation in
function sqlite3BtreeOpen() seems to ignore any setting of
sqlite3SharedCacheEnabled() if it is a in-memory database. What about having
in-memory databases named like: ":memory:". So as to be
able to indicate to sqlite3_open_v2() to create multiple in-memory db handlers
sharing the same btree.
Actually after a quick and dirty hack in btree.c's sqlite3BtreeOpen()
things _seem_ to work correctly. Of course this is not in any way how it should
be done, but I didn't see a quick way to give in-memory databases a name so
that can be used in the search for existing in-memory databases. But it
indicates that the btree can be shared also for in-memory databases.
//FIXME-XXX originally this is isMemdb=0
if( (flags & BTREE_PRIVATE)==0
&& isMemdb==1
//FIXME-XXX originally this is strcmp(zFulPathname, ...
if( 0==strcmp("", sqlite3PagerFilename(pBt->pPager))
&& sqlite3PagerVfs(pBt->pPager)==pVfs ){
What do you think? I think I could manage to correctly hack the sqlite code so
that it provides naming for :memory: databases. Of course only if the basic
sharing of memory databases between multiple db handles is supposed to work.
Markus
_________________________________________________________________
Invite your mail contacts to join your friends list with Windows Live Spaces.
It's easy!
http://spaces.live.com/spacesapi.aspx?wx_action=create&wx_url=/friends.aspx&mkt=en-us
_______________________________________________
sqlite-users mailing list
[email protected]
http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
