Re: ec.saveChages gives null exception (didn't before) SOLVED
Re: ec.saveChages gives null exception (didn't before) SOLVED
- Subject: Re: ec.saveChages gives null exception (didn't before) SOLVED
- From: Drew Thoeni <email@hidden>
- Date: Tue, 8 Mar 2005 20:58:43 -0500
As I was typing the below, making sure I'd covered all the bases, I
went to verify that each of the things you all had pointed out were
actually working. Pierre wins the prize!
This table is large so some columns were beyond the scroll window and
one of them was, in fact, missing the external type (must have hit the
delete key at the wrong time). Since I kept looking at the PK column, I
didn't see this. Simple enough. But hard to find.
I'm leaving the below text in hope that the thread will help someone
else in the future. Back to coding!
regards and thanks again.
Drew
---
First, thank you all for the tips. I'm learning a lot because of this
particular problem.
Now, more news...
I've done some more experimenting and some very odd things are
occurring. The facts:
1) This behavior only happens on inserts and updates to this table
(classroom)
2) I was able to, not sure how, coax it into adding a single record.
The odd part, very odd, was the PK assigned by Frontbase was 2147483647
3) This number, interestingly is 1 + the precision limit of the
database.
4) Subsequent inserts fail, not at the insert statement, but now at the
"SELECT UNIQUE" statement, which is trying to obtain a PK from
Frontbase
5) I have reset the PK generator in Frontbase and tested the "SELECT
UNIQUE" statement in the database. It works fine, both before and after
resetting the PK.
6) All columns have ...ooops... HERE THE PROBLEM.
So, given this, it appears there is some failure in EOModel to
accurately request (or receive) the PK for this table only.
On Mar 8, 2005, at 1:14 PM, Chuck Hill wrote:
On Mar 8, 2005, at 3:52 AM, Drew Thoeni wrote:
I dropped the jdbc2info and allowed EOM to recreate it. Still, the
problem persists. I also posted again in response to an off list
response I got regarding exposure of the PK.
I've traced the SQL and have a related question. Just prior to EOF
asking the database for the primary key (to prepare for an insert)
EOF does a call to the database selecting *every* record in the table
about to be updated. This is trivial for a test database, but in
production, if you had, say, 100,000 records, it seems like this is
unneeded and burdensome behavior from EOF.
Why does EOF do this and should it be stopped?
It sounds like some form of pessimistic locking. What does your JDBC
URL look like? Check the docs for the FB JDBC driver. You want to
have this on the end of the JDBC URL:
/isolation=read_committed/locking=optimistic
OK, that said, the JDBC URL is probably not the cause of the select.
Does it do this for every table, or just some? It is certainly not
_supposed_ to be doing this.
Chuck
On Mar 7, 2005, at 11:28 PM, Chuck Hill wrote:
On Mar 7, 2005, at 5:46 PM, Drew Thoeni wrote:
I've researched Chuck's idea (below) and EOModeler jdbc2info says
I'm using the right info (FrontBase). FYI, I'm not planning on
changing databases frequently.
I also followed up on Arturo's suggestion and my foreign keys are
marked as "NO" under "Deferrable" in "Check" in the table
definition. So, a couple of questions:
While this is a good idea, I doubt it is the cause of the exception
message below. The message below suggests that the error in either
in the jdbc2info or in how you have modeled an attribute. Double
check your model. Then check it again.
1) Can the "deferred" value be set in EOModeler and passed to the
DB via the SQL command in EOModeler?
No, this is not a function of EOModeler.
2) If not, (and this is really a Frontbase question) it appears you
can't modify this in Frontbase once it is set (must delete
referential rule and readd). Is this correct. Frontbase user's
guide is silent on this matter.
I believe that is correct. That is what I have always done in any
case.
Chuck
Regards,
Drew
On Mar 7, 2005, at 12:37 PM, Chuck Hill wrote:
It is more likely what you have not done. When you connect to a
database in EOModeler, it caches some database specific
information to the connection dictionary under the jdbc2info key.
If you later change databases (as I recall you have) and don't use
EOModeler to connect to the new database, this jdbc2info will
still be for the old database. I'm not 100% sure that connecting
to the new DB will update this, it's not something I do
frequently.
If you inspect the root node in the EOModel you should see this
jdbc2info key. Delete it and save the model. If you want you can
then connect to the new database and save the model, or EOF can
regenerate this information from the database each time it is run.
If you are going to switch back and forth a lot, I would try to
avoid saving this information in the model. Note that if it is
not in the model, EOF will keep two connections to the database
open, one for data and one for this information. In most cases
this does not matter, but it will in you are in a environment
where the DB license is on a per connection basis.
Chuck
On Mar 6, 2005, at 8:08 PM, Drew Thoeni wrote:
For quite some time, this method, and all others that save to the
Class table (like a classroom, not an object definition) worked
fine. I've looked at the SQL that is being generated and it
appears fine. There is only one not-null column in the table (the
key) and the failures are on updates to an existing record.
Clearly, something is null that should not be. And, it seems it's
a result of something I've done. But I've spent about an hour
tinkering with this and (probably in 10 minutes) reached my level
of expertise.
Can anyone suggest what I might check into regarding this
exception?
[2005-03-06 22:49:59 EST] <main> Waiting for requests...
[2005-03-06 22:58:35 EST] <WorkerThread8>
java.lang.NullPointerException
at
com.webobjects.jdbcadaptor.JDBCAdaptor.isValidQualifierType(JDBCAd
aptor.java:659)
at
com.webobjects.eoaccess.EODatabaseContext.isValidQualifierTypeForA
ttribute(EODatabaseContext.java:5146)
at
com.webobjects.eoaccess.EODatabaseContext.lockingNonQualifiableAtt
ributes(EODatabaseContext.java:5160)
at
com.webobjects.eoaccess.EODatabaseContext.createAdaptorOperationsF
orDatabaseOperationAttributes(EODatabaseContext.java:5346)
at
com.webobjects.eoaccess.EODatabaseContext.createAdaptorOperationsF
orDatabaseOperation(EODatabaseContext.java:5520)
at
com.webobjects.eoaccess.EODatabaseContext.performChanges(EODatabas
eContext.java:6337)
at
com.webobjects.eocontrol.EOObjectStoreCoordinator.saveChangesInEdi
tingContext(EOObjectStoreCoordinator.java:411)
at
com.webobjects.eocontrol.EOEditingContext.saveChanges(EOEditingCon
text.java:3148)
at ClassAdmin.saveClass(ClassAdmin.java:63)
--
Practical WebObjects - a book for intermediate WebObjects
developers who want to increase their overall knowledge of
WebObjects, or those who are trying to solve specific application
development problems.
http://www.global-village.net/products/practical_webobjects
--
Practical WebObjects - a book for intermediate WebObjects developers
who want to increase their overall knowledge of WebObjects, or those
who are trying to solve specific application development problems.
http://www.global-village.net/products/practical_webobjects
--
Practical WebObjects - a book for intermediate WebObjects developers
who want to increase their overall knowledge of WebObjects, or those
who are trying to solve specific application development problems.
http://www.global-village.net/products/practical_webobjects
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden