That's the last of the RDBMS migration steps done! Yay! Note that the code currently has some unit testing issues: - Metakit needs some attention in a couple of areas - RDBMS backends are having trouble ordering their journal entries correctly. I'm going to be migrating them to use TIMESTAMP for the date column, but that's not necessarily going to fix things as mysql and postgresql both appear to have second granularity. Sqlite will ignore the data type as usual ;) Next up is the datatype-ification of the RDBMS backends. Part of that will involve the migration to numeric IDs, which will also be done in the *dbm backends (already done in metakit). The ID exposed *above* the hyperdb will be String, since so many things assume a string ID now.

--- a/TODO.txt	Fri Mar 19 05:27:55 2004 +0000
+++ b/TODO.txt	Sun Mar 21 23:39:08 2004 +0000
@@ -1,10 +1,9 @@
 This file has been re-purposed to contain specifically the items that need
 doing before the next release:
 
-- indexing in RDBMSes
-- add tests for group-by-multilink so I finally implement it for the RDBMSes
-- s/getnode/getitem in backends (and s/Node/Item)
 - have rdbms backends look up the journal for actor if it's not set
-- ensure index creation is triggered by the version 1->2 update
+- migrate rdbms backends to use typed columns
+- migrate to numeric ID values (fixes bug 817217)
 
-- migrate to numeric ID values (fixes bug 817217)
+- ensure index creation is triggered by the version 1->2 update
+  (and other upgrade tests)