I have a lot of things I want to mention, so let's just get right to it.
Roller should bundle Moblogger
It'd be cool if Roller bundled Russ's Moblogger as part of the application. It'd be pretty easy, all we'd need to do is slap a UI on it's config file - possibly put the config into a database, and maybe even use Roller's "page" feature for the e-mail templates. Then we'd have to figure out a way to start Moblogger's thread in the background and possibly do one thread for each user.
Roller Plugins
Should Roller add a Plugins feature like Blojsom has? Why should we? If we should, maybe we should just re-use an existing plug-in API - i.e. Struts' or Blojsom's. What would make a good plugin? Is this just a fancy name for a feature that can be turned on/off (i.e. comments).
Struts Menu for the Editor
I hope to add the ability to choose different menu types in Roller's Editor UI. First, I'll have to integrate Roller's tabbed menu as a displayer for struts-menu [demo]. The CoolMenu's displayer will work great (and require less mouse clicks). I mention this in hopes of making Lance and Dave aware of any schema changes that need to be made to support a user-selectable menu type. I'll also need to add the ability to specify struts-forwards in the menu-config.xml (not implemented yet in struts-menu).
This site and my provider
This site was down a lot today - probably because of all the hits from javablogs.com (just put Hibernate in your title). I noticed that Simon and Sam's sites were down as well. This leads me to believe that my kgbinternet.com is not up to snuff. Dave and Lance seem to never have issues with Roller crashing - so why am I? Please recommend any hosting providers that offer private Tomcat instances (preferrably with lots of RAM and disk space).
What's up with Wrox and Professional JSP 2.0
Who knows... I haven't heard a thing, from anyone. Think I'd get anyone's attention if I slapped both of my chapters on this site - free of charge? I doubt it. I might just do that if I don't get any feedback in the next week or so. Are they gonna sue me for the pocket change they paid me to write the chapters?
Now I'm going to assign this to my Java category and see if all you java.bloggers can crash this site again. ;-)
My last upgrade from Hibernate 1.2.x to 2.0.x was a little rocky, so I'm going to document it again today - this time with a Hibernate 2.0-enabled version of XDoclet, and code samples from last time. Today I'm upgrading from 1.2.4 to 2.0 rc1.
1. First, I downloaded Hibernate 2.0 rc1. BTW, I also updated to the latest version of XDoclet from CVS.
2. Next, I copied it to my project's lib directory, deleted bin, doc, and src directories. The only files I need are hibernate2.jar and (in the lib folder) c3p0.jar, cglib.jar, dom4j.jar, jcs.jar, jdom.jar and odmg.jar. One thing to note is with Hibernate 1.2.x, I didn't need dom4j.jar, jdom.jar or cglib.jar in my deployed WEB-INF/lib folder. However, with the new version, I did need to include them.
3. Then I updated lib/lib.properties and began changing my code with gool ol' Homesite.
4. Hey, step 4 is the same as last time! Edit build.xml file to pick up the new DTD. I changed <hibernate/> to be <hibernate validatexml="true" version="2.0"/>.
5. Using HomeSite, I did s/cirrus.hibernate/net.sf.hibernate/g for all my .java and .xml files. Don't forget to change your log4.properties and hibernate.properties files.
6. In hibernate.cfg.xml I changed cirrus.hibernate.sql.OracleDialect to net.sf.hibernate.dialect.OracleDialect. I also updated the DTD to 2.0 and moved all properties inside the <session-factory> element. Here's a sample hibernate.cfg.xml file from struts-resume.
7. In my ServiceLocator.java class, I changed the following from (this is so JUnit Tests can run w/o JNDI):
Datastore datastore = Hibernate.createDatastore();
datastore.storeClass(package.MyClass.class); // lots of these
sf = datastore.buildSessionFactory();
To:
sf = new Configuration().addClass(MyClass.class) // lots of these
.buildSessionFactory();
8. Time to bring Eclipse to the table. I refreshed the project and added the new hibernate2.jar to the "Java Build Path."
9. In my StartupListener.java (which initializes Hibernate for Tomcat), I changed the following code:
// Configure Hibernate.
try {
Hibernate.configure();
if (log.isDebugEnabled()) {
log.debug("Hibernate configuration completed.");
}
} catch (HibernateException h) {
log.fatal("Error configuring Hibernate!", h);
}
To:
try {
SessionFactory sf =
new Configuration().configure().buildSessionFactory();
if (log.isDebugEnabled()) {
log.debug("Hibernate configuration completed.");
}
} catch (HibernateException h) {
log.fatal("Error configuring Hibernate!", h);
}
10. Changed all instances of readonly="true" to inverse="false" inverse="true" for XDoclet tags. Also changed role to name.
11. Removed length attribute from any <key-property ... /> elements in my .hbm.xml files. These were rudely put there by the ReverseGenerator. Not all my mapping files are generated by XDoclet, if I have one with a composite-id, I just create the mapping file manually. I also had the change the DTD manually in these files.
12. In my build.xml, I changed the SchemaExport class from net.sf.hibernate.tools to net.sf.hibernate.tool.hbm2ddl.
13. OK, here goes, my first attempt at compiling (honestly!). Darn - I forgot to save StartupListener.java - trying again (this time with "ant clean" first). Sweeeet - it worked!
14. Hmmm, some log messages I wasn't expecting. I forgot to change my log4j.properties file from cirrus.hibernate to net.sf.hibernate. I'll add this above (step 5) to make this how to more accurate.
15. Running persistence tests - since I keep my Oracle driver (ojdbc14.jar) in ${hibernate.dir}/lib, I had to move it to the new directory. Now this is a strange error:
[junit] 2003-04-29 13:14:50,968 INFO [main] Configuration.addClass(264) | Mapping resource: com/
.../persistence/ChangeRequest.hbm.xml
[junit] 2003-04-29 13:14:51,234 ERROR [main] XMLHelper$ErrorLogger.error(37) | Error parsing XML
: XML InputStream(19)
[junit] org.xml.sax.SAXParseException: Attribute "name" is required and must be specified for el
ement type "param".
My guess is that it's being caused by:
<generator class="sequence">
<param>cr_master_sq</param>
</generator>
After reading the Hibernate2 Porting Guidelines, looks like I need to add a name attribute. Didn't the exception tell me that! ;0)
I'll update this post when I figure out the answer to my question on Hibernate's forums...
Update: I had to 1) upgrade my XDoclet JARs, and 2) replace the following XDoclet tags:
* @hibernate.id column="cr_id" unsaved-value="null"
* generator-class="sequence" generator-parameter-1="cr_master_sq"
with:
* @hibernate.id column="cr_id" unsaved-value="null" generator-class="sequence"
* @hibernate.generator-param name="sequence" value="cr_master_sq"
Now I'm experiencing this issue.
Update 2: I discovered (from Gavin) that readonly="true" is the same as inverse="true". I've updated step 10 above, and now my upgrade is complete!
Whaddya know, I write a popular post on javablogs.com and this site ends up crashing for a few hours. Here's the cause:
Unexpected Signal : 11 occurred at PC=0x403B4304
Function=(null)+0x403B4304
Library=/usr/java/j2sdk1.4.1_01/jre/lib/i386/client/libjvm.so
NOTE: We are unable to locate the function name symbol for the error
just occurred. Please refer to release documentation for possible
reason and solutions.
Lovely...
I've been using CVS for a couple of years now, but only recently have I been responsible for tagging and branching. Dave is using the branching feature on Roller, and we're using it at our day job. Here's how I think it works - please correct me if I'm wrong.
When you create a branch, you can continue working on that branch - providing you check it out. This means that you cannot have HEAD checked out and update from the branch. Or can you? I tried this with Roller a couple night ago, and I don't think I got the 091_BRANCH. However, last night when I tried to check stuff in, it said I was on a branch. Wierd.
At work right now, I've created a 2.0 branch, and we're working on this to fix bugs in our beta release. The other day, I checked out HEAD, and did a merge from the 2.0 branch. Then I created a 2.0.1 branch. When we're with 2.0 and we make it a final release, I'll do another merge with HEAD. Is this how you do it? Do you work on a branch and merge/re-tag every few days? It'd be cool if checking in a file on a branch automagically checked it into the HEAD.
In the manuscript I'm reading, I just ran into a diagram and overview of a layered architecture - and all of a sudden, it hit me. At work, most of our business rules are being implemented by Struts' Validator framework. In the case of indexed properties and more complicated rules, these reside in the validate() method of our action forms. So it's interesting to me that I'm using a business delegate to perform my business logic - when in actuality, all my delegates are doing is copying properties from a POJO to an ValidatorForm. So this begs the question - shouldn't validation (a.k.a business rules) be done wherever it is most convenient rather than only in the domain layer? Oh yeah, and most of this is just validation - our real business rules (comparing data for validation) takes place in Oracle stored procedures. And you know what - it works great!. So my opinion is - do whatever is easiest and makes the most sense.
