|
From: SourceForge.net <no...@so...> - 2006-11-14 12:04:45
|
Bugs item #1592786, was opened at 2006-11-08 17:20 Message generated for change (Comment added) made by jayhem You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1592786&group_id=74338 Please note that this message will contain a full copy of the comment thread, including the initial issue submission, for this request, not just the latest update. Category: SQLite Group: None Status: Open Resolution: Invalid Priority: 5 Private: No Submitted By: Jayhem (jayhem) Assigned to: Nobody/Anonymous (nobody) Summary: so-sqlite data corruption during updates Initial Comment: I'm hitting a bug when using SO with turbogears 1.1b with an sqlite database. When I do several successive updates to a table (which are very slow btw), data gets corrupted in the interface, select() returns invalid objects, as if objects were pulled from a cache that was out of synch with the db. >From this point, successive select() without any updates in-between can return different values, randomly selecting one of the previsouly set values. I have created a simple test-case (on turbogears) Just launch it and change some values. Usually after 10-20 changes, the corruption appears. I have tried this same testcase with MySQL and do not hit this problem. Note that dev.cfg declares sqlobject.dburi="notrans_sqlite:///..." file: model.py ====================================================== from sqlobject import * from turbogears.database import PackageHub hub = PackageHub("bug") __connection__ = hub class Item(SQLObject): data = UnicodeCol(length=40) def loaddb(): hub.begin() Item.dropTable(ifExists=True, cascade=True) hub.commit() Item.createTable(ifNotExists=True) for i in range(20): Item(data = str(i)) hub.commit() loaddb() file: controllers.py ====================================================== import turbogears from turbogears.controllers import RootController, expose from model import Item class Root(RootController): @expose(template="bug.templates.index") def index(self): items = Item.select() return dict(items = items) @expose(template="bug.templates.index") def change(self, id): item = Item.get(id) item.data += str(id) items = Item.select() return dict(items = items) file: index.kid ====================================================== <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"> <html xmlns="http://www.w3.org/1999/xhtml" xmlns:py="http://purl.org/kid/ns#" py:extends="'master.kid'"> <head> <meta content="text/html; charset=UTF-8" http-equiv="content-type" py:replace="''"/> <title>Items</title> </head> <body> <ul py:for="i in items"> <li>${i.data} <a href="/change/${i.id}">[edit]</a></li> </ul> </body> </html> ---------------------------------------------------------------------- >Comment By: Jayhem (jayhem) Date: 2006-11-14 12:04 Message: Logged In: YES user_id=1640652 You are right. the bug does not seem to cause data corruption in the database (i.e it's not a db related issue), it just causes random results to be returned from queries. Simply switching the database to MySQL solves the problem, which confirms the bug is not in the db. An even more annoying issue, which I think is related, is the extremely slow inserts/updates into sqlite, and frequent error messages telling me the db is locked. I am new to turbogears, sqlobject, sqlite and python, so I'm sorry I won't have time to work on a SqlObject-only testcase. Let me know if I can do anything to help fix this. ---------------------------------------------------------------------- Comment By: Oleg Broytmann (phd) Date: 2006-11-13 16:59 Message: Logged In: YES user_id=4799 Our company has been using SQLObject with SQLite for three rather large commercial programs. Neither of our developers, testers or users ever experienced data loss. Are you sure the problem is not TG-related? Please write an SQLObject-only test. In any case I cannot run your program - I don't have TG and not going to install it. ---------------------------------------------------------------------- You can respond by visiting: https://sourceforge.net/tracker/?func=detail&atid=540672&aid=1592786&group_id=74338 |
