I have had problems with pooling in the past; currently I have an OJB
based app running (against a MySQL db) using the following pooling:
Vincent Frison wrote:
Hi all,
I have sometimes troubles with pooling. One of the most frequent exception is:
An action has failed: Borrow broker from pool fa
I now have a running version, but the way Thomas proposed would be much
nicer.
I have added a collection descriptor
and he returns both addresses.
I tried to modify my getters and setters for addresses to set
homeAddress and workAddress, but I tested this and noticed that they are
not invo
i tried to change my code to using extends, it works perfect for me, and
from an object oriented design view, this is perfect!
what i did (if i apply the things i did to georgs problem):
- create an abstract class Address that has all attributes in it
- derive both classes HomeAddress and WorkAdd
Even if it has been already post, i just want to clarify the tip :
- according to Oracle 9i notes
http://www.oracle.com/technology/tech/java/sqlj_jdbc/htdocs/jdbc_faq.htm#08_01
it's important to note that mapping date attributes such as
java.sql.Time java.util.Date (with hours ...) and even
jav
Thomas Dudziak wrote:
As far as I understand he wants to use a reference that refers to a
class with a compound primarykey, where one of the foreignkey values
is fixed:
Yes, that is what I am searching for.
-
To unsubscr
As far as I understand he wants to use a reference that refers to a
class with a compound primarykey, where one of the foreignkey values
is fixed:
That would actually be a neat feature to have IMO ?!
Tom
-
To unsubsc
Hi all,
I have sometimes troubles with pooling. One of the most frequent exception is:
An action has failed: Borrow broker from pool failed, using PBKey
org.apache.ojb.broker.PBKey: jcdAlias=default, user=null, password=null
It seems that I explode the broker pool capacity (altough I close broke
Yes, you need to declare getters and setters, at least for the fields
that are used for primarykeys and foreignkeys.
But the benefit is that you can put whatever implementation you see
fit, the do not even need to have actual java fields for the values.
And the speed hit because of the reflection i
Hi,
not sure that I understand the problem.
Each User has two Address attributes ('home' and 'work'), both are of
the same type 'Address' (both 1:1 references), you don't use extension
classes AddressHome and AddressWork objects. Address use a composite PK
('login' and 'type').
In this case eac
hi,
as far as i understand this, you try to differentiate by the value of
"adress.type" whether an adress is a home wor work adress?
well, i have no solution for that, but i tried it here with a similair data
structur, and it didn't work either.
to be honest i'm not sure if this is the intended us
Sorry for that second posting. I sent this before I subscribed to the
list, so I thought it will not be posted (now, with a latency of nearly
one day)
Regards,
Georg
-
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional com
yes you are right - its solution
but unfortunatelly it's not convenient to me ;(
i have 71 beans now and i'm using xdoclet - so when I' ll decide replace
GroupImpl to GroupAnotherImpl i'll have much work
but its better then nothing - thanks ;)
if I want find another solution I'll use yours
about
Maksimenko Alexander wrote:
but...
according to dtd i should write either extent or field-descripto
elements -
in the second part a "extent-class*" declaration is defined too. The
first part mean that at least an extent-class definition is needed.
Armin
its dtd from 1.0.1
Armin Waibel wrote:
but...
according to dtd i should write either extent or field-descripto
elements -
its dtd from 1.0.1
Armin Waibel wrote:
I saw ReferenceTest test worked with OJB 1.0.1 too, with a different
mapping of the interface. Did you tried to declare the
reference-descriptor and fk-field in the interf
Hi,
I have an object "User" who has 2 addresses, home and work.
User is mapped to a table user (no problem here).
Both home and work address are mapped to a table address.
This table has a composite primary key: login and type
- login from table user as a foreign key
- type is a string (or enum)
The proxies comment is irrelevant to the solution I noted.
You need to change:
To:
As far as my comment regarding proxies, in general we get better
performance by NOT proxying reference descriptors. We do however
increase performance by proxying collection descriptors. I
but in this case i need create setters in interface, isn't it ?
Thomas Dudziak wrote:
I was wondering why you map the Impl classes at all ? You can declare
the fields for the interface, you only need to change the
PersistentFieldClass in OJB.properties to
PersistentFieldIntrospectorImpl or Persiste
thanks for idea but its unfortunately doesn't fit to me :(
because i'm need using proxies for perfomance reasons
of cause user a 1-m to group - but i created this example to simplify
understanding by problem - in real project my classes are big enough and
ojb descriptor is huge ;)
Gelhar, Wallac
I saw ReferenceTest test worked with OJB 1.0.1 too, with a different
mapping of the interface. Did you tried to declare the
reference-descriptor and fk-field in the interface class mapping too?
e.g.
Class RefObject is an interface
Armin Waibel wrote:
Maksimenko Alexander wrote:
I'm using 1.0.1 version
ojb doesnt produce exception but it generates
SELECT A0.GROUP_ID,A0.ID FROM USER A0 WHERE A0.GROUP_ID = ?
instead of:
SELECT A0.GROUP_ID,A0.ID FROM USER A0, group1 a1 WHERE A0.GROUP_ID =
a1.id and a1.manager_id=?
AFAIK Jakob
I was wondering why you map the Impl classes at all ? You can declare
the fields for the interface, you only need to change the
PersistentFieldClass in OJB.properties to
PersistentFieldIntrospectorImpl or PersistentFieldAutoProxyImpl.
I use it this way all the time (with factory-class/factory-metho
I've ran into this before. I think the solution I found was to map the
reference / collection descriptor classes to the implementing descriptor
as follows:
Also just note that it usually doesn't pay
Maksimenko Alexander wrote:
I'm using 1.0.1 version
ojb doesnt produce exception but it generates
SELECT A0.GROUP_ID,A0.ID FROM USER A0 WHERE A0.GROUP_ID = ?
instead of:
SELECT A0.GROUP_ID,A0.ID FROM USER A0, group1 a1 WHERE A0.GROUP_ID =
a1.id and a1.manager_id=?
AFAIK Jakob fixed some bugs in c
I'm using 1.0.1 version
ojb doesnt produce exception but it generates
SELECT A0.GROUP_ID,A0.ID FROM USER A0 WHERE A0.GROUP_ID = ?
instead of:
SELECT A0.GROUP_ID,A0.ID FROM USER A0, group1 a1 WHERE A0.GROUP_ID =
a1.id and a1.manager_id=?
Armin Waibel wrote:
Hi,
your mapping and query seems ok. Cou
Hi,
your mapping and query seems ok. Could you post the stack trace? Did you
try latest from CVS OJB 1.0.x branch (branch OJB_1_0_RELEASE)?
In OJB test-suite we have a similar test
ReferenceTest#testDeepPathQuery(), this test pass.
regards,
Armin
Maksimenko Alexander wrote:
hi!
it's wondering bu
Thanks for answer but it doesnt solve the problem:
ojb generates
SELECT A0.GROUP_ID,A0.ID FROM USER A0 WHERE A0.GROUP_ID = ?
instead of:
SELECT A0.GROUP_ID,A0.ID FROM USER A0, group1 a1 WHERE A0.GROUP_ID =
a1.id and a1.manager_id=?
[EMAIL PROTECTED] wrote:
You should map your DB and make the quer
You should map your DB and make the queries with the xxxImpl classes.
In your case :
QueryByCriteria query = QueryFactory.newQuery(UserImpl.class,new
Criteria());
and
is useless.
David WIESZTAL.
-Original Message-
From: Maksimenko Alexander [mailto:[EMAIL PROTECTED]
Se
hi!
it's wondering but I found out that its difficult to use interface
driven design with ojb ;(
lets consider a simple 2-table example:
interface User{Integer getId(),Group getGroup()},
class UserImpl implements User{private Integer id; private Group group;
access methods}
interface Group{Int
Hi,
I just saw another post about Oracle 9i and Blobs, so if somebody is going to
get a look into it, here is some more:
When using the Oracle9i platform storing Blobs/Clobs does not work when using
the DBCP connection factory
(org.apache.ojb.broker.accesslayer.ConnectionFactoryDBCPImpl)
This is
29 matches
Mail list logo