Changes between Version 2 and Version 3 of TracSprint

Timestamp:

Mar 7, 2008, 9:54:36 AM (16 years ago)

Author:

Christian Boos

Comment:

Added some caveats for SQLAlchemy

TracSprint

@@ -9 +9 @@
  * give evil_twin a brain
  * Work on sqlalchemy branch
+   ,,(but see caveats belove -- cboos),,
  * Internationalization
  * Pluggable user-directory provider
@@ -18 +19 @@
  * Port [http://trac-hacks.org/wiki/TracBlogPlugin TracBlogPlugin] to 0.11
  * Work on Tags based permission policy plugin
+
+
+----
+== Notes ==
+
+=== SQLAlchemy ===
+I'd hate to see a lot of effort spent on the ''superficial'' aspect of the SQLAlchemy port, because that could be eventually wasted if we don't ''first'' answer the more fundamental questions. 
+The data model creation part has been done (I think), we know porting the statements is doable (only lots of work) - what we don't know yet is if the benefits brought by SQLAlchemy on the infrastructure outweights the disadvantage of adding yet another required dependency.
+
+So I'd welcome any progress on the SQLAlchemy integration, but it has to address the right questions:
+ - better pool handling? 
+ - no new / different multithreading issues? 
+ - evaluate the impact of SQLAlchemy on all the open issues for
+   [query:?status=!closed&keywords=~postgres PostgreSQL],
+   [query:?status=!closed&keywords=~sqlite SQLite],
+   [query:?status=!closed&keywords=~mysql MySQL]
+   i.e. does SQLAlchemy help, make things worse, is indifferent to the issue?
+
+  
+