Just a followup, it could be that the frontpage lists old entries because it can only find one user with a handle and so is only listing those users. If you hit it in the next hour or so, http://rr.javalobby.org:20900 should be up for testing. It also appears that hitting the site from 'links' on linux redirects properly, while firefox does not. As I said, a hard redirect to main.do is probably not a good thing and unless I changed Jroller from the default, I don't think 1.2 did that :)

-Matt

Matthew Schmidt wrote:
Hey guys. We're in the process of trying to migrate to Roller 2.0 and also ran into some of the indices problems. Easily resolved, but there are other issues.

1) Just visiting / doesn't appear to redirect to main.do or forward to main.jsp like it used to. We'd like to not have a hard redirect to main.do as that would screw with our page ranks and indexing with the various search engines.
2) If I do go to /, I get the following error:

javax.servlet.jsp.JspException: Exception; TEMPLATE_CLASSPATH=/menu-tabbed.vm; exception=org.roller.RollerPermissionsException: Only global admin or system
user can save properties
at org.roller.presentation.tags.VelocityTag.doEndTag(VelocityTag.java:101)
    at org.roller.presentation.tags.HybridTag.doEndTag(HybridTag.java:64)
at _jsp._theme._tiles_22dfrontpage__jsp._jspService(theme/tiles-frontpage.jsp:45)


3) Hitting main.do gives me a bunch of old entries, back to 2003.

4) The hot blogs list has all blogs listed with a /page as the url, no handle. It appears that none of the blogs have a handle after the upgrade. There also doesn't appear to be anything in the upgrade script that generates them. 5) If I login as one of the users, it doesn't think I have any weblogs, when I clearly did before.

6) In roller.properties, you may want to allow system property expansion for the planet cache directory.

Thanks guys, looking forward to trying out the new release my our data.

-Matt

Matt Raible wrote:
On 11/14/05, Anil Gangolli <[EMAIL PROTECTED]> wrote:
I retested my upgrade and I didn't see these errors.  Is it possible
that you had arrived at your 1.3 database schema via a development build
of the schema that didn't include all of the indices?

This happened to me on two different indices.

ERROR 1091 (42000): Can't DROP 'pingqueueentry_pingtargetid_index';
check that column/key exists
ERROR 1091 (42000): Can't DROP 'pingqueueentry_websiteid_index'; check
that column/key exists

I tracked my errors back to the fact that I had arrived at my 1.3 schema
via a 1.2 dev schema, not actually the 1.2 release schema; late in the
game in 1.2, some additional indices were added. I think this happened
just before 1.0 or 1.1 as well.  Anyway, the 1.3 to 2.0 upgrade didn't
find these when dropping them, but they are indeed present if built from
the official released versions.

It's entirely possible (and highly likely) that my current 1.2 schema
is from development, rather than a release.  These indices must have
been added in 1.2 (or prior) b/c I did upgrade 1.2 to 1.3 and then to
2.0.

No worries, I'll just remember to comment out these indices when
running the upgrade script.

Matt

--a.


Anil Gangolli wrote:

Typically DDL (schema modifying commands) are not transactional.
There is no way to commit versus roll them back.  So yes, it is
especially important that they work properly with no errors.  If
errors are skipped during sourcing (which I think is the default in
MySQL) and if objects being dropped already do not exist, that should
be ok.  However, this might cause problems with other databases.

Matt Raible wrote:

I ended up having to comment out 4 of the indexes b/c they didn't
exist in my database.  Here's the list from the MySQL
130-to-200-migration script.

--alter table webpage drop index webpage_id_index;
--alter table website drop index website_id_index;
--alter table folderassoc drop index folderassoc_relation_index;
--alter table weblogentry drop index weblogentry_pinnedtomain_index;

I think it's important that we remove these (or add an "if exists")
b/c the migration scripts seems to leave the database in an unknown
state when this fails (rather than rolling back all the changes).

Matt


On 11/13/05, Matt Raible <[EMAIL PROTECTED]> wrote:


When trying to upgrade from 1.3 to 2.0 using the
130-to-200-migration.sql script, I got an error:

ERROR 1091 (42000) at line 111: Can't DROP 'webpage_id_index'; check
that column/key exists

Any ideas?

Matt

On 11/13/05, Matt Raible <[EMAIL PROTECTED]> wrote:


I should mention my prod and test environments:

Production on Linux:

MySQL 3.23.56-Max-log with mysql-connector-java-3.1.10-bin.jar
Tomcat 5.0.28
JDK 5

Testing on OS X:

MySQL 5.0.15 with mysql-connector-java-3.1.7-bin.jar
Tomcat 5.5.9
JDK 5

Matt

On 11/13/05, Matt Raible <[EMAIL PROTECTED]> wrote:


I'm testing an upgrade of my site from 1.2 to 1.3, as well as 2.0
tonight. I'm just finishing up the 1.3 upgrade and I've only found
one issue.

On raibledesigns.com, if you scroll to the bottom of the page, there
will be a link on the left side that allows you to navigate to the
previous entry.  It seems like it's a bug in 1.2 that I can only
navigate to the single previous entry, rather than a list of entries. In 1.3, there is no link at all. Did this functionality get removed
in 1.3?

Also, comparing my raibledesigns.com blog to my jroller.com blog, I
noticed there are no navigation links to go to next/prev/main when
viewing a single entry.

I was able to fix this by adding in "#showNextPreviousLinks()" to my theme. However, my main page is _sunsets and is not displayed as a menu item - but it does show up as the name to navigate back to from
the #showNextPreviousLinks() macro.

On to 1.3 to 2.0, then 1.2 to 2.0. Should have the results in less
than an hour.

Matt

On 11/13/05, Dave Johnson <[EMAIL PROTECTED]> wrote:


Thanks to our testers, over the weekend we've found and fixed these:

* removed session-factory name argument from Hibernate configuration
file
* restored missing create-user page
* restored missing weblog maintenance page
* added cache flush to save of bookmarks
* fixed array-index out of bounds problem when no default blogroll
defined

Are we ready for a new RC, or is there some other evil lurking?

- Dave





Reply via email to