headers
Armin Waibel | 1 Feb 2005 16:21
Picon
Favicon

Re: OJB 1.0.2 ?

hi all,

will check in the fixes for OJB 1.0.2 later. I have a touch (or more)
of fever. I have fixed one test case in odmg-api and found problems with 
cache handling.
Sorry folks.

regards,
Armin

Armin Waibel wrote:
> Think I found a bug.
> I got these output
>  [junit] Running org.apache.ojb.broker.AllTests
>  [junit] Tests run: 517, Failures: 1, Errors: 0, Time elapsed: 142,359 sec
>  [junit] Test org.apache.ojb.broker.AllTests FAILED
>  [junit] Running org.apache.ojb.odmg.AllTests
>  [junit] [BOOT] INFO: Found logging properties file: OJB-logging.properties
>  [junit] [BOOT] INFO: Logging: Found logger class 
> 'org.apache.ojb.broker.util.logging.PoorMansLoggerImpl
>  [junit] [BOOT] INFO: Loading OJB's properties from file 
> file:/E:/dev/ojb_1.0.x/db-ojb/target/test/ojb/OJB.properties
>  [junit] Tests run: 129, Failures: 2, Errors: 0, Time elapsed: 15,828 sec
> 
> The failure in PB tests was caused by an odmg-api test. If I run this 
> test standalone with the emptyCacheImpl the test pass, with default and 
> twoLevel cache this test fails. If I rollback ObjectEnvelopeTable to a 
> version before the new reordering, the test pass with default cache. 
> Seems to be a problem in reordering or in modification state detection.
>
(Continue reading)

Permalink | Reply |
headers
Brian McCallister | 1 Feb 2005 16:31
Favicon

Re: OJB 1.0.2 ?

Hope you feel better soon!

-Brian

On Feb 1, 2005, at 10:21 AM, Armin Waibel wrote:

> hi all,
>
> will check in the fixes for OJB 1.0.2 later. I have a touch (or more)
> of fever. I have fixed one test case in odmg-api and found problems 
> with cache handling.
> Sorry folks.
>
> regards,
> Armin
>
> Armin Waibel wrote:
>> Think I found a bug.
>> I got these output
>>  [junit] Running org.apache.ojb.broker.AllTests
>>  [junit] Tests run: 517, Failures: 1, Errors: 0, Time elapsed: 
>> 142,359 sec
>>  [junit] Test org.apache.ojb.broker.AllTests FAILED
>>  [junit] Running org.apache.ojb.odmg.AllTests
>>  [junit] [BOOT] INFO: Found logging properties file: 
>> OJB-logging.properties
>>  [junit] [BOOT] INFO: Logging: Found logger class 
>> 'org.apache.ojb.broker.util.logging.PoorMansLoggerImpl
>>  [junit] [BOOT] INFO: Loading OJB's properties from file 
>> file:/E:/dev/ojb_1.0.x/db-ojb/target/test/ojb/OJB.properties
(Continue reading)

Permalink | Reply |
headers
Brian McCallister | 4 Feb 2005 19:07
Favicon

Re: OJB 1.0.2 ?

Was chatting with former coworker (Hi Terry!) who observed that 1.0.2  
is *not* drop-in compatible with 1.0.1.

I sort of feel that it either needs to be (with deprecations is fine),  
or needs to be 1.1 in order to indicate this.

-Brian

On Feb 1, 2005, at 10:31 AM, Brian McCallister wrote:

> Hope you feel better soon!
>
> -Brian
>
> On Feb 1, 2005, at 10:21 AM, Armin Waibel wrote:
>
>> hi all,
>>
>> will check in the fixes for OJB 1.0.2 later. I have a touch (or more)
>> of fever. I have fixed one test case in odmg-api and found problems  
>> with cache handling.
>> Sorry folks.
>>
>> regards,
>> Armin
>>
>> Armin Waibel wrote:
>>> Think I found a bug.
>>> I got these output
>>>  [junit] Running org.apache.ojb.broker.AllTests
(Continue reading)

Permalink | Reply |
headers
Thomas Dudziak | 4 Feb 2005 22:58
Picon
Gravatar

Re: OJB 1.0.2 ?

Could you be more specific ? Which public API parts have changed ?

Tom
Permalink | Reply |
headers
Armin Waibel | 5 Feb 2005 00:25
Picon
Favicon

Re: OJB 1.0.2 ?

In CVS 1.0.x branch changes (I know) are:

- Locking was moved from odmg-implementation to OJB kernel. Users using 
own odmg.locking.LockMap implementations have to adapt implementations 
to ...broker.locking.LockMap interface

- In ObjectCache interface I introduced a new method. Users using own 
caching implementations have introduce the new method.
But it seems that I can move this new method to an internal interface, 
so that ObjectCache doesn't need to change.
Will check in changes soon.

regards,
Armin

Thomas Dudziak wrote:
> Could you be more specific ? Which public API parts have changed ?
> 
> Tom
> 
> ---------------------------------------------------------------------
> To unsubscribe, e-mail: ojb-dev-unsubscribe <at> db.apache.org
> For additional commands, e-mail: ojb-dev-help <at> db.apache.org
> 
> 
>
Permalink | Reply |
headers
Brian McCallister | 5 Feb 2005 20:43
Favicon

Re: OJB 1.0.2 ?

The locking stuff is the big one, I haven't tried dropping it into a 
big app as (sadly) I am not using it in a big app at my new job (yet 
;-).

We may want to create a delegating, deprecated, lock map with the old 
name in the 1.0.X and leave it out of 1.1

-Brian

On Feb 4, 2005, at 6:25 PM, Armin Waibel wrote:

> In CVS 1.0.x branch changes (I know) are:
>
> - Locking was moved from odmg-implementation to OJB kernel. Users 
> using own odmg.locking.LockMap implementations have to adapt 
> implementations to ...broker.locking.LockMap interface
>
> - In ObjectCache interface I introduced a new method. Users using own 
> caching implementations have introduce the new method.
> But it seems that I can move this new method to an internal interface, 
> so that ObjectCache doesn't need to change.
> Will check in changes soon.
>
> regards,
> Armin
>
> Thomas Dudziak wrote:
>> Could you be more specific ? Which public API parts have changed ?
>> Tom
>> ---------------------------------------------------------------------
(Continue reading)

Permalink | Reply |
headers
Armin Waibel | 7 Feb 2005 21:51
Picon
Favicon

Re: OJB 1.0.2 ?

Brian McCallister wrote:
> The locking stuff is the big one, I haven't tried dropping it into a big 
> app as (sadly) I am not using it in a big app at my new job (yet ;-).
>
> We may want to create a delegating, deprecated, lock map with the old 
> name in the 1.0.X and leave it out of 1.1
>

hmm, I will try to restore the 'old' odmg-locking stuff (delegating is 
not possible), try to simplify new stuff and make both locking api 
useable (the old locking api can cause write lock concurrency problems, 
maybe own LockMap implementations fix this issue).

Armin

> -Brian
> 
> On Feb 4, 2005, at 6:25 PM, Armin Waibel wrote:
> 
>> In CVS 1.0.x branch changes (I know) are:
>>
>> - Locking was moved from odmg-implementation to OJB kernel. Users 
>> using own odmg.locking.LockMap implementations have to adapt 
>> implementations to ...broker.locking.LockMap interface
>>
>> - In ObjectCache interface I introduced a new method. Users using own 
>> caching implementations have introduce the new method.
>> But it seems that I can move this new method to an internal interface, 
>> so that ObjectCache doesn't need to change.
>> Will check in changes soon.
(Continue reading)

Permalink | Reply |
headers
Brian McCallister | 7 Feb 2005 22:10
Favicon

Re: OJB 1.0.2 ?

I think delegating to new is probably preferable, just with old class  
name.

-Brian

On Feb 7, 2005, at 3:51 PM, Armin Waibel wrote:

> Brian McCallister wrote:
>> The locking stuff is the big one, I haven't tried dropping it into a  
>> big app as (sadly) I am not using it in a big app at my new job (yet  
>> ;-).
>>
>> We may want to create a delegating, deprecated, lock map with the old  
>> name in the 1.0.X and leave it out of 1.1
>>
>
> hmm, I will try to restore the 'old' odmg-locking stuff (delegating is  
> not possible), try to simplify new stuff and make both locking api  
> useable (the old locking api can cause write lock concurrency  
> problems, maybe own LockMap implementations fix this issue).
>
> Armin
>
>
>> -Brian
>> On Feb 4, 2005, at 6:25 PM, Armin Waibel wrote:
>>> In CVS 1.0.x branch changes (I know) are:
>>>
>>> - Locking was moved from odmg-implementation to OJB kernel. Users  
>>> using own odmg.locking.LockMap implementations have to adapt
(Continue reading)

Permalink | Reply |

Gmane