You are viewing a plain text version of this content. The canonical link for it is here.
Posted to ojb-dev@db.apache.org by Brian McCallister <br...@apache.org> on 2004/04/05 21:31:49 UTC
OTM CollectionProxy Bug =(
On the otm-dependent="true" collections, in the otm unit test mappings,
add a proxy="true" and bad things start happening =( (ie, dependent
tests all fail).
I took a stab at it but the solution wasn't obvious. Oleg, any
insights? This is a bit of a big problem for me at the moment.
-Brian
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <br...@apache.org>.
Hmm, last email blocked by sending with wrong email addy =(
Okay, more info. Looks like if I get an explicit write lock everything
works fine, the OTMCollectionProxyListener is set on the correct
collection, but when read lock and implicit locking is on (so should be
able to freely write), the listener is set on a collection which is
never materialized.
I am trying to trace through the various pieces of copying code and am
deeply suspicious of the copy-by-serialization that CollectionProxy
instances go through (via SimpleKit) as they are passed to a reflective
copier but don't have a no-arg constructor.
-Brian
On Apr 5, 2004, at 3:57 PM, Brian McCallister wrote:
> are NOT deleted that is
>
> -Brian
>
> On Apr 5, 2004, at 3:42 PM, Brian McCallister wrote:
>
>> Follow that up by changing the refs to the interface so that it *can*
>> proxy =)
>>
>> otm-dependent objects are deleted =(
>>
>> -Brian
>>
>> Patch to replicate bug:
>>
>> Index: src/test/org/apache/ojb/repository_junit_otm.xml
>> ===================================================================
>> RCS file:
>> /home/cvs/db-ojb/src/test/org/apache/ojb/repository_junit_otm.xml,v
>> retrieving revision 1.5
>> diff -u -r1.5 repository_junit_otm.xml
>> --- src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 13:58:59 -0000 1.5
>> +++ src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 19:41:00 -0000
>> @@ -58,6 +58,7 @@
>> element-class-ref="org.apache.ojb.otm.AddressDesc"
>>
>> collection-
>> class="org.apache.ojb.broker.util.collections.ManageableArrayList"
>> otm-dependent="true"
>> + proxy="true"
>> >
>> <inverse-foreignkey field-ref="personId"/>
>> </collection-descriptor>
>> Index: src/test/org/apache/ojb/otm/Person.java
>> ===================================================================
>> RCS file: /home/cvs/db-ojb/src/test/org/apache/ojb/otm/Person.java,v
>> retrieving revision 1.3
>> diff -u -r1.3 Person.java
>> --- src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 12:16:25
>> -0000 1.3
>> +++ src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 19:41:00 -0000
>> @@ -16,6 +16,7 @@
>> */
>>
>> import java.util.ArrayList;
>> +import java.util.List;
>> import java.io.Serializable;
>>
>> public class Person implements Serializable
>> @@ -26,7 +27,7 @@
>> private String lastname;
>> private Integer mainAddressId;
>> private Address mainAddress;
>> - private ArrayList otherAddresses;
>> + private List otherAddresses;
>>
>> public Person()
>> {
>> @@ -88,7 +89,7 @@
>> this.mainAddress = mainAddress;
>> }
>>
>> - public ArrayList getOtherAddresses()
>> + public List getOtherAddresses()
>> {
>> return otherAddresses;
>> }
>>
>>
>>
>> On Apr 5, 2004, at 3:31 PM, Brian McCallister wrote:
>>
>>> On the otm-dependent="true" collections, in the otm unit test
>>> mappings, add a proxy="true" and bad things start happening =( (ie,
>>> dependent tests all fail).
>>>
>>> I took a stab at it but the solution wasn't obvious. Oleg, any
>>> insights? This is a bit of a big problem for me at the moment.
>>>
>>> -Brian
>>>
>>>
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>>
>>>
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
Seems the OTMCollectionProxyListener#afterLoad is never being called.
The load is called before the listener is added, but I cannot seem to
find the right place to use it =( Inserts are identified, but not
deletes.
-Brian
On Apr 5, 2004, at 3:57 PM, Brian McCallister wrote:
> are NOT deleted that is
>
> -Brian
>
> On Apr 5, 2004, at 3:42 PM, Brian McCallister wrote:
>
>> Follow that up by changing the refs to the interface so that it *can*
>> proxy =)
>>
>> otm-dependent objects are deleted =(
>>
>> -Brian
>>
>> Patch to replicate bug:
>>
>> Index: src/test/org/apache/ojb/repository_junit_otm.xml
>> ===================================================================
>> RCS file:
>> /home/cvs/db-ojb/src/test/org/apache/ojb/repository_junit_otm.xml,v
>> retrieving revision 1.5
>> diff -u -r1.5 repository_junit_otm.xml
>> --- src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 13:58:59 -0000 1.5
>> +++ src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 19:41:00 -0000
>> @@ -58,6 +58,7 @@
>> element-class-ref="org.apache.ojb.otm.AddressDesc"
>>
>> collection-
>> class="org.apache.ojb.broker.util.collections.ManageableArrayList"
>> otm-dependent="true"
>> + proxy="true"
>> >
>> <inverse-foreignkey field-ref="personId"/>
>> </collection-descriptor>
>> Index: src/test/org/apache/ojb/otm/Person.java
>> ===================================================================
>> RCS file: /home/cvs/db-ojb/src/test/org/apache/ojb/otm/Person.java,v
>> retrieving revision 1.3
>> diff -u -r1.3 Person.java
>> --- src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 12:16:25
>> -0000 1.3
>> +++ src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 19:41:00 -0000
>> @@ -16,6 +16,7 @@
>> */
>>
>> import java.util.ArrayList;
>> +import java.util.List;
>> import java.io.Serializable;
>>
>> public class Person implements Serializable
>> @@ -26,7 +27,7 @@
>> private String lastname;
>> private Integer mainAddressId;
>> private Address mainAddress;
>> - private ArrayList otherAddresses;
>> + private List otherAddresses;
>>
>> public Person()
>> {
>> @@ -88,7 +89,7 @@
>> this.mainAddress = mainAddress;
>> }
>>
>> - public ArrayList getOtherAddresses()
>> + public List getOtherAddresses()
>> {
>> return otherAddresses;
>> }
>>
>>
>>
>> On Apr 5, 2004, at 3:31 PM, Brian McCallister wrote:
>>
>>> On the otm-dependent="true" collections, in the otm unit test
>>> mappings, add a proxy="true" and bad things start happening =( (ie,
>>> dependent tests all fail).
>>>
>>> I took a stab at it but the solution wasn't obvious. Oleg, any
>>> insights? This is a bit of a big problem for me at the moment.
>>>
>>> -Brian
>>>
>>>
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>>
>>>
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
Update: Works if you obtain an explicit write lock on the parent --
seems it doesn't respect implicit locking =(
-Brian
On Apr 5, 2004, at 3:57 PM, Brian McCallister wrote:
> are NOT deleted that is
>
> -Brian
>
> On Apr 5, 2004, at 3:42 PM, Brian McCallister wrote:
>
>> Follow that up by changing the refs to the interface so that it *can*
>> proxy =)
>>
>> otm-dependent objects are deleted =(
>>
>> -Brian
>>
>> Patch to replicate bug:
>>
>> Index: src/test/org/apache/ojb/repository_junit_otm.xml
>> ===================================================================
>> RCS file:
>> /home/cvs/db-ojb/src/test/org/apache/ojb/repository_junit_otm.xml,v
>> retrieving revision 1.5
>> diff -u -r1.5 repository_junit_otm.xml
>> --- src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 13:58:59 -0000 1.5
>> +++ src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
>> 19:41:00 -0000
>> @@ -58,6 +58,7 @@
>> element-class-ref="org.apache.ojb.otm.AddressDesc"
>>
>> collection-
>> class="org.apache.ojb.broker.util.collections.ManageableArrayList"
>> otm-dependent="true"
>> + proxy="true"
>> >
>> <inverse-foreignkey field-ref="personId"/>
>> </collection-descriptor>
>> Index: src/test/org/apache/ojb/otm/Person.java
>> ===================================================================
>> RCS file: /home/cvs/db-ojb/src/test/org/apache/ojb/otm/Person.java,v
>> retrieving revision 1.3
>> diff -u -r1.3 Person.java
>> --- src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 12:16:25
>> -0000 1.3
>> +++ src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 19:41:00 -0000
>> @@ -16,6 +16,7 @@
>> */
>>
>> import java.util.ArrayList;
>> +import java.util.List;
>> import java.io.Serializable;
>>
>> public class Person implements Serializable
>> @@ -26,7 +27,7 @@
>> private String lastname;
>> private Integer mainAddressId;
>> private Address mainAddress;
>> - private ArrayList otherAddresses;
>> + private List otherAddresses;
>>
>> public Person()
>> {
>> @@ -88,7 +89,7 @@
>> this.mainAddress = mainAddress;
>> }
>>
>> - public ArrayList getOtherAddresses()
>> + public List getOtherAddresses()
>> {
>> return otherAddresses;
>> }
>>
>>
>>
>> On Apr 5, 2004, at 3:31 PM, Brian McCallister wrote:
>>
>>> On the otm-dependent="true" collections, in the otm unit test
>>> mappings, add a proxy="true" and bad things start happening =( (ie,
>>> dependent tests all fail).
>>>
>>> I took a stab at it but the solution wasn't obvious. Oleg, any
>>> insights? This is a bit of a big problem for me at the moment.
>>>
>>> -Brian
>>>
>>>
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>>
>>>
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <br...@apache.org>.
are NOT deleted that is
-Brian
On Apr 5, 2004, at 3:42 PM, Brian McCallister wrote:
> Follow that up by changing the refs to the interface so that it *can*
> proxy =)
>
> otm-dependent objects are deleted =(
>
> -Brian
>
> Patch to replicate bug:
>
> Index: src/test/org/apache/ojb/repository_junit_otm.xml
> ===================================================================
> RCS file:
> /home/cvs/db-ojb/src/test/org/apache/ojb/repository_junit_otm.xml,v
> retrieving revision 1.5
> diff -u -r1.5 repository_junit_otm.xml
> --- src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
> 13:58:59 -0000 1.5
> +++ src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
> 19:41:00 -0000
> @@ -58,6 +58,7 @@
> element-class-ref="org.apache.ojb.otm.AddressDesc"
>
> collection-
> class="org.apache.ojb.broker.util.collections.ManageableArrayList"
> otm-dependent="true"
> + proxy="true"
> >
> <inverse-foreignkey field-ref="personId"/>
> </collection-descriptor>
> Index: src/test/org/apache/ojb/otm/Person.java
> ===================================================================
> RCS file: /home/cvs/db-ojb/src/test/org/apache/ojb/otm/Person.java,v
> retrieving revision 1.3
> diff -u -r1.3 Person.java
> --- src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 12:16:25
> -0000 1.3
> +++ src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 19:41:00 -0000
> @@ -16,6 +16,7 @@
> */
>
> import java.util.ArrayList;
> +import java.util.List;
> import java.io.Serializable;
>
> public class Person implements Serializable
> @@ -26,7 +27,7 @@
> private String lastname;
> private Integer mainAddressId;
> private Address mainAddress;
> - private ArrayList otherAddresses;
> + private List otherAddresses;
>
> public Person()
> {
> @@ -88,7 +89,7 @@
> this.mainAddress = mainAddress;
> }
>
> - public ArrayList getOtherAddresses()
> + public List getOtherAddresses()
> {
> return otherAddresses;
> }
>
>
>
> On Apr 5, 2004, at 3:31 PM, Brian McCallister wrote:
>
>> On the otm-dependent="true" collections, in the otm unit test
>> mappings, add a proxy="true" and bad things start happening =( (ie,
>> dependent tests all fail).
>>
>> I took a stab at it but the solution wasn't obvious. Oleg, any
>> insights? This is a bit of a big problem for me at the moment.
>>
>> -Brian
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <br...@apache.org>.
Follow that up by changing the refs to the interface so that it *can*
proxy =)
otm-dependent objects are deleted =(
-Brian
Patch to replicate bug:
Index: src/test/org/apache/ojb/repository_junit_otm.xml
===================================================================
RCS file:
/home/cvs/db-ojb/src/test/org/apache/ojb/repository_junit_otm.xml,v
retrieving revision 1.5
diff -u -r1.5 repository_junit_otm.xml
--- src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
13:58:59 -0000 1.5
+++ src/test/org/apache/ojb/repository_junit_otm.xml 5 Apr 2004
19:41:00 -0000
@@ -58,6 +58,7 @@
element-class-ref="org.apache.ojb.otm.AddressDesc"
collection-
class="org.apache.ojb.broker.util.collections.ManageableArrayList"
otm-dependent="true"
+ proxy="true"
>
<inverse-foreignkey field-ref="personId"/>
</collection-descriptor>
Index: src/test/org/apache/ojb/otm/Person.java
===================================================================
RCS file: /home/cvs/db-ojb/src/test/org/apache/ojb/otm/Person.java,v
retrieving revision 1.3
diff -u -r1.3 Person.java
--- src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 12:16:25
-0000 1.3
+++ src/test/org/apache/ojb/otm/Person.java 5 Apr 2004 19:41:00 -0000
@@ -16,6 +16,7 @@
*/
import java.util.ArrayList;
+import java.util.List;
import java.io.Serializable;
public class Person implements Serializable
@@ -26,7 +27,7 @@
private String lastname;
private Integer mainAddressId;
private Address mainAddress;
- private ArrayList otherAddresses;
+ private List otherAddresses;
public Person()
{
@@ -88,7 +89,7 @@
this.mainAddress = mainAddress;
}
- public ArrayList getOtherAddresses()
+ public List getOtherAddresses()
{
return otherAddresses;
}
On Apr 5, 2004, at 3:31 PM, Brian McCallister wrote:
> On the otm-dependent="true" collections, in the otm unit test
> mappings, add a proxy="true" and bad things start happening =( (ie,
> dependent tests all fail).
>
> I took a stab at it but the solution wasn't obvious. Oleg, any
> insights? This is a bit of a big problem for me at the moment.
>
> -Brian
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
No problem =)
-Brian
On Apr 7, 2004, at 6:45 PM, Oleg Nitz wrote:
> Hi Brian,
>
> Yes, I am radically changing ConcreteEditingContext, so please don't
> commit
> your fix to CVS if it is possible.
>
> Regards,
> Oleg
>
> On Wednesday 07 April 2004 16:48, Brian McCallister wrote:
>> Ack!
>>
>> I have been digging into it and am right next to having it solved, I
>> think (as I know what is causing it, finally). Is a trixy thing to
>> make
>> sure nothing else is broken by fix though =(
>>
>> Short form: OTMCollectionProxyListener is being set on a swizzlded
>> copy
>> of the Collectionproxy which is actually given back to the client. The
>> client loads the one it is given, but the swizzled copy is never
>> laoded.
>>
>> The listener populates the stored values in the entry/context
>> (initialized to null before proxy is loaded) -- except it never
>> actually gets to do so.
>>
>> When it comes time to compare for deletions the original collection is
>> null so the comparator presumes nothing has changed.
>>
>> Are you radically changing ConcreteEditingContext (it needs some
>> refactoring)? If so I will hold off on fixing it. if not, will work on
>> getting it fixed this morning as this is basically breaking our app
>> here (as workaround is to not proxy otm-dependent collections, and
>> OutOfMemory errors are annoying) =(
>>
>> -Brian
>>
>> On Apr 7, 2004, at 3:11 AM, Oleg Nitz wrote:
>>> Hi Brian,
>>>
>>> I'll look into this problem soon. Now I'm finishing OTM refactoring.
>>> I am sorry very much for the delay in your work that I cause.
>>>
>>> Oleg
>>>
>>> On Monday 05 April 2004 22:31, Brian McCallister wrote:
>>>> On the otm-dependent="true" collections, in the otm unit test
>>>> mappings,
>>>> add a proxy="true" and bad things start happening =( (ie, dependent
>>>> tests all fail).
>>>>
>>>> I took a stab at it but the solution wasn't obvious. Oleg, any
>>>> insights? This is a bit of a big problem for me at the moment.
>>>>
>>>> -Brian
>>>>
>>>>
>>>>
>>>>
>>>> --------------------------------------------------------------------
>>>> -
>>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
On Apr 8, 2004, at 8:52 AM, Armin Waibel wrote:
>> For 1.0, however, we could call the OTM unstable (adopt Cocoon
>> semantics).
>
> +1 think this is a honest paraphrase ;-)
Sorry, didn't catch that this is what you were saying =)
-Brian
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Armin Waibel <ar...@apache.org>.
Brian McCallister wrote:
> For 1.0, however, we could call the OTM unstable (adopt Cocoon semantics).
>
+1 think this is a honest paraphrase ;-)
Further on we will need some changes in otm package for 1.1, e.g. the
object copy strategy stuff should be moved to the kernel, because we
need the object copy stuff in the new two-level cache.
The otm.transaction package is duplicated code from odmg.transaction,
this should be moved to kernel too, because PB-api use this classes too.
And we need a lot more of OTM documentation (similar to Brian's
introduction), how locking and relations work, pitfalls, performance
optimations,...
regards,
Armin
> -Brian
>
> On Apr 8, 2004, at 8:35 AM, Brian McCallister wrote:
>
>> I thought we decided back in December to "official support" the OTM in
>> 1.0. I am not completely sure it should be as the code is a little
>> less than mature/stable/fully functional right now -- but Oleg is
>> working hard, and I'm trying hard to find the bugs =)
>>
>> That said, I think we *need* to support the OTM as it is the only way
>> outside of ODMG to get automatic dirtying, etc. "Outside of ODMG" is
>> important as the ODMG is all but undocumented (you can find lots of
>> ODMG docs, but it requires google, and the docs are more often on O2
>> which is awfully similar at least).
>>
>> -Brian
>>
>> On Apr 8, 2004, at 8:06 AM, Armin Waibel wrote:
>>
>>> Hi Robert,
>>>
>>> Robert S. Sfeir wrote:
>>>
>>>> Can't help but cringe when I see that code is still getting changed
>>>> in RC mode :-) I know I know It's busted and it will be better, I
>>>> just think that when future released are planned, we ought to at
>>>> least have a rule that says if we go RC no new features or full
>>>> rewrites of stuff, just bug fixes.
>>>>
>>>
>>> AFAIK the OTM-api is not official supported in 1.0 - but maybe I'm
>>> wrong. In that case continual development is allowed in my opinion ;-)
>>>
>>> regards,
>>> Armin
>>>
>>>> JMHO
>>>> R
>>>> ---------------------------------------------------------------------
>>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>>
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>>
>>>
>>>
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
For 1.0, however, we could call the OTM unstable (adopt Cocoon
semantics).
-Brian
On Apr 8, 2004, at 8:35 AM, Brian McCallister wrote:
> I thought we decided back in December to "official support" the OTM in
> 1.0. I am not completely sure it should be as the code is a little
> less than mature/stable/fully functional right now -- but Oleg is
> working hard, and I'm trying hard to find the bugs =)
>
> That said, I think we *need* to support the OTM as it is the only way
> outside of ODMG to get automatic dirtying, etc. "Outside of ODMG" is
> important as the ODMG is all but undocumented (you can find lots of
> ODMG docs, but it requires google, and the docs are more often on O2
> which is awfully similar at least).
>
> -Brian
>
> On Apr 8, 2004, at 8:06 AM, Armin Waibel wrote:
>
>> Hi Robert,
>>
>> Robert S. Sfeir wrote:
>>
>>> Can't help but cringe when I see that code is still getting changed
>>> in RC mode :-) I know I know It's busted and it will be better, I
>>> just think that when future released are planned, we ought to at
>>> least have a rule that says if we go RC no new features or full
>>> rewrites of stuff, just bug fixes.
>>>
>>
>> AFAIK the OTM-api is not official supported in 1.0 - but maybe I'm
>> wrong. In that case continual development is allowed in my opinion
>> ;-)
>>
>> regards,
>> Armin
>>
>>> JMHO
>>> R
>>> ---------------------------------------------------------------------
>>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>>
>>
>>
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
I thought we decided back in December to "official support" the OTM in
1.0. I am not completely sure it should be as the code is a little less
than mature/stable/fully functional right now -- but Oleg is working
hard, and I'm trying hard to find the bugs =)
That said, I think we *need* to support the OTM as it is the only way
outside of ODMG to get automatic dirtying, etc. "Outside of ODMG" is
important as the ODMG is all but undocumented (you can find lots of
ODMG docs, but it requires google, and the docs are more often on O2
which is awfully similar at least).
-Brian
On Apr 8, 2004, at 8:06 AM, Armin Waibel wrote:
> Hi Robert,
>
> Robert S. Sfeir wrote:
>
>> Can't help but cringe when I see that code is still getting changed
>> in RC mode :-) I know I know It's busted and it will be better, I
>> just think that when future released are planned, we ought to at
>> least have a rule that says if we go RC no new features or full
>> rewrites of stuff, just bug fixes.
>>
>
> AFAIK the OTM-api is not official supported in 1.0 - but maybe I'm
> wrong. In that case continual development is allowed in my opinion ;-)
>
> regards,
> Armin
>
>> JMHO
>> R
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Armin Waibel <ar...@apache.org>.
Hi Robert,
Robert S. Sfeir wrote:
> Can't help but cringe when I see that code is still getting changed in
> RC mode :-) I know I know It's busted and it will be better, I just
> think that when future released are planned, we ought to at least have a
> rule that says if we go RC no new features or full rewrites of stuff,
> just bug fixes.
>
AFAIK the OTM-api is not official supported in 1.0 - but maybe I'm
wrong. In that case continual development is allowed in my opinion ;-)
regards,
Armin
> JMHO
>
> R
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by "Robert S. Sfeir" <ro...@codepuccino.com>.
Oleg Nitz wrote:
>Hi Brian,
>
>Yes, I am radically changing ConcreteEditingContext, so please don't commit
>your fix to CVS if it is possible.
>
Can't help but cringe when I see that code is still getting changed in
RC mode :-) I know I know It's busted and it will be better, I just
think that when future released are planned, we ought to at least have a
rule that says if we go RC no new features or full rewrites of stuff,
just bug fixes.
JMHO
R
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Oleg Nitz <on...@ukr.net>.
Hi Brian,
Yes, I am radically changing ConcreteEditingContext, so please don't commit
your fix to CVS if it is possible.
Regards,
Oleg
On Wednesday 07 April 2004 16:48, Brian McCallister wrote:
> Ack!
>
> I have been digging into it and am right next to having it solved, I
> think (as I know what is causing it, finally). Is a trixy thing to make
> sure nothing else is broken by fix though =(
>
> Short form: OTMCollectionProxyListener is being set on a swizzlded copy
> of the Collectionproxy which is actually given back to the client. The
> client loads the one it is given, but the swizzled copy is never
> laoded.
>
> The listener populates the stored values in the entry/context
> (initialized to null before proxy is loaded) -- except it never
> actually gets to do so.
>
> When it comes time to compare for deletions the original collection is
> null so the comparator presumes nothing has changed.
>
> Are you radically changing ConcreteEditingContext (it needs some
> refactoring)? If so I will hold off on fixing it. if not, will work on
> getting it fixed this morning as this is basically breaking our app
> here (as workaround is to not proxy otm-dependent collections, and
> OutOfMemory errors are annoying) =(
>
> -Brian
>
> On Apr 7, 2004, at 3:11 AM, Oleg Nitz wrote:
> > Hi Brian,
> >
> > I'll look into this problem soon. Now I'm finishing OTM refactoring.
> > I am sorry very much for the delay in your work that I cause.
> >
> > Oleg
> >
> > On Monday 05 April 2004 22:31, Brian McCallister wrote:
> >> On the otm-dependent="true" collections, in the otm unit test
> >> mappings,
> >> add a proxy="true" and bad things start happening =( (ie, dependent
> >> tests all fail).
> >>
> >> I took a stab at it but the solution wasn't obvious. Oleg, any
> >> insights? This is a bit of a big problem for me at the moment.
> >>
> >> -Brian
> >>
> >>
> >>
> >>
> >> ---------------------------------------------------------------------
> >> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> >> For additional commands, e-mail: ojb-dev-help@db.apache.org
> >
> > ---------------------------------------------------------------------
> > To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> > For additional commands, e-mail: ojb-dev-help@db.apache.org
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <br...@apache.org>.
Ack!
I have been digging into it and am right next to having it solved, I
think (as I know what is causing it, finally). Is a trixy thing to make
sure nothing else is broken by fix though =(
Short form: OTMCollectionProxyListener is being set on a swizzlded copy
of the Collectionproxy which is actually given back to the client. The
client loads the one it is given, but the swizzled copy is never
laoded.
The listener populates the stored values in the entry/context
(initialized to null before proxy is loaded) -- except it never
actually gets to do so.
When it comes time to compare for deletions the original collection is
null so the comparator presumes nothing has changed.
Are you radically changing ConcreteEditingContext (it needs some
refactoring)? If so I will hold off on fixing it. if not, will work on
getting it fixed this morning as this is basically breaking our app
here (as workaround is to not proxy otm-dependent collections, and
OutOfMemory errors are annoying) =(
-Brian
On Apr 7, 2004, at 3:11 AM, Oleg Nitz wrote:
> Hi Brian,
>
> I'll look into this problem soon. Now I'm finishing OTM refactoring.
> I am sorry very much for the delay in your work that I cause.
>
> Oleg
>
> On Monday 05 April 2004 22:31, Brian McCallister wrote:
>> On the otm-dependent="true" collections, in the otm unit test
>> mappings,
>> add a proxy="true" and bad things start happening =( (ie, dependent
>> tests all fail).
>>
>> I took a stab at it but the solution wasn't obvious. Oleg, any
>> insights? This is a bit of a big problem for me at the moment.
>>
>> -Brian
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Brian McCallister <mc...@forthillcompany.com>.
Ack!
I have been digging into it and am right next to having it solved, I
think (as I know what is causing it, finally).
Short form: OTMCollectionProxyListener is being set on a swizzlded copy
of the Collectionproxy which is actually given back to the client. The
client loads the one it is given, but the swizzled copy is never
laoded.
The listener populates the stored values in the entry/context
(initialized to null before proxy is loaded) -- except it never
actually gets to do so.
When it comes time to compare for deletions the original collection is
null so the comparator presumes nothing has changed.
Are you radically changing ConcreteEditingContext (it needs some
refactoring)? If so I will hold off on fixing it. if not, will work on
getting it fixed this morning as this is basically breaking our app
here (as workaround is to not proxy otm-dependent collections, and
OutOfMemory errors are annoying) =(
-Brian
On Apr 7, 2004, at 3:11 AM, Oleg Nitz wrote:
> Hi Brian,
>
> I'll look into this problem soon. Now I'm finishing OTM refactoring.
> I am sorry very much for the delay in your work that I cause.
>
> Oleg
>
> On Monday 05 April 2004 22:31, Brian McCallister wrote:
>> On the otm-dependent="true" collections, in the otm unit test
>> mappings,
>> add a proxy="true" and bad things start happening =( (ie, dependent
>> tests all fail).
>>
>> I took a stab at it but the solution wasn't obvious. Oleg, any
>> insights? This is a bit of a big problem for me at the moment.
>>
>> -Brian
>>
>>
>>
>>
>> ---------------------------------------------------------------------
>> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
>> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
>
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org
Re: OTM CollectionProxy Bug =(
Posted by Oleg Nitz <on...@ukr.net>.
Hi Brian,
I'll look into this problem soon. Now I'm finishing OTM refactoring.
I am sorry very much for the delay in your work that I cause.
Oleg
On Monday 05 April 2004 22:31, Brian McCallister wrote:
> On the otm-dependent="true" collections, in the otm unit test mappings,
> add a proxy="true" and bad things start happening =( (ie, dependent
> tests all fail).
>
> I took a stab at it but the solution wasn't obvious. Oleg, any
> insights? This is a bit of a big problem for me at the moment.
>
> -Brian
>
>
>
>
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
> For additional commands, e-mail: ojb-dev-help@db.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: ojb-dev-unsubscribe@db.apache.org
For additional commands, e-mail: ojb-dev-help@db.apache.org