Yes Ralf,
this is exactly what I need!
The possibility to optionally use the caching feature from the classes generated by SourceGen.
This could be done in two ways:
1) Modifying SourceGen to optionally get generated classes implementing the NOT static unmarshal function.
2) Modifying the static marshal function code so that it optionally implements caching.
Massimo Pasina
-----Messaggio originale-----
Da: Ralf Joachim [mailto:[EMAIL PROTECTED]]
Inviato: martedì 14 marzo 2006 17.15
A: [email protected]
Oggetto: Re: [castor-user] R: [castor-user] Caching unmarshall
descriptors with Castor's Source Code Generato r
As far as I understand Massimo he does not have problems with
redeploying his application. He only likes to have the caching feature
implemented inside castor or generated by SourceGen inside the descriptors.
Having said that I also don't know how the container handles
redeployment of applications.
Ralf
Marco Mistroni schrieb:
> Hello ralf,
> Thx.. had a quick look
> Frankly I have never used that feature... nor had a look at code
> To what it does
>
> I have few question then:
>
> 1 - will that 'caching' behaviour of Unmarshaller survive application
> redeployment? (I must admit that I don't know how classloader works when
> Redeploying apps)
> 2 - if redeployment of changed app means server restart, then I suppose
> Massimo won't have any problems of having in cache 'old' versions and
> New versions..
>
> Regards
> marco
>
> -----Original Message-----
> From: Ralf Joachim [mailto:[EMAIL PROTECTED]]
> Sent: 14 March 2006 15:22
> To: [email protected]
> Subject: Re: [castor-user] R: [castor-user] Caching unmarshall descriptors
> with Castor's Source Code Generato r
>
> Marco, Werner,
>
> to get an idea about what Massimo does, take a look at:
>
> http://castor.codehaus.org/xml-faq.html#How-can-I-speed-up-marshalling/unmar
> shalling-performance?
>
> Similar to Werner, I'd also suggest to create a new issue in jira first.
> As far as I understand you have implemented a ClassDescriptorResolver.
> It should not be too difficult for you to create an testcase that
> unmarshalls a simple XML with and without your ClassDescriptorResolver
> and attach it to this issue.
>
> Having said that I had not enough time to take a look at that yet.
>
> Ralf
>
>
> Marco Mistroni schrieb:
>
>>Hello,
>>
>>
>>>Because if we regenerate the classes with a new XML TAG, it wouldn't be
>>
>>cached any longer.
>>
>>Does this assume a server restart?
>>
>>If not, does at least involve an ‘application restart’? cos that is what
>>normally happens when you deploy/redeploy
>>
>>In that case, you’ll have a new cache when app is redeployed…. Correct?
>>
>>
>>
>>HTH
>>
>> marco
>>
>>
>>
>>
>>
>>
>>
>>------------------------------------------------------------------------
>>
>>*From:* Pasina Massimo [mailto:[EMAIL PROTECTED]]
>>*Sent:* 14 March 2006 11:21
>>*To:* [email protected]
>>*Cc:* [EMAIL PROTECTED]; Nicotra Leopoldo
>>*Subject:* [castor-user] R: [castor-user] Caching unmarshall descriptors
>>with Castor's Source Code Generato r
>>
>>
>>
>>Dear Werner,
>>we don't like the solution we have made, because it makes static the
>>methods implemented in Castor library. We consider it only as a
>>temporary patch to solve a problem that could have been solved in a much
>>better way. Because if we regenerate the classes with a new XML TAG, it
>>wouldn't be cached any longer.
>>
>>So we don't think it's useful to incorporate this fix into a new release
>>of Castor.
>>We think the best solution would be to change the generation mechanism,
>>in order to have the application server to reload the new generated
>
> classes.
>
>>But this has still to be implemented.
>>Sincerely,
>>Massimo Pasina & Leopoldo Nicotra
>>
>>-----Messaggio originale-----
>>Da: Werner Guttmann [mailto:[EMAIL PROTECTED]]
>>Inviato: lunedì 13 marzo 2006 21.33
>>A: [email protected]
>>Oggetto: Re: [castor-user] Caching unmarshall descriptors with Castor's
>>Source Code Generato r
>>
>>
>>
>>Buena sera Massimo,
>>
>>given you have already implemented the desired functionality, why not
>>create a new feature request at http://jira.codehaus.org/browse/CASTOR,
>>and attach the proposed solution to the newly created Jira issue as a
>>unified patch ? I am sure somebody will review your code and come back
>>to you with comments.
>>
>>Thanks
>>Werner Guttmann
>>
>>Pasina Massimo wrote:
>>
>>>Hi, we are a software house that develops web applications using Java and
>>>
>>>Xml. We use some application server for testing our applications and in
>>>
>>>particular IBM Websphere Application Server 5.1 and Tomcat 4.
>>>
>>>
>>>
>>>We have some perfomance problem with Oracle Application Server (10 times
>>>
>>>slower) in marshalling and unmarshalling the standard class generatated
>
> by
>
>>>SourceGenerator (with no caching).
>>>
>>>
>>>
>>>We read the FAQ on your site and we have modified the classes in order to
>>>
>>>cache descriptors for unmarshal, according to the FAQ, and so we got the
>>>
>>>OracleAS perfomance similar to Tomcat.
>>>
>>>
>>>
>>>Now the questions are:
>>>
>>>
>>>
>>>* Why the generated classes are not automatically cached by your package?
>>>
>>>
>>>
>>>* Is it possible to use the SourceGenerator task in order to generate
>>>
>>>the classes with descriptor caching?
>>>
>>>
>>>
>>>* Is it possible for us to modify the SourceGenerator's package in
>>>
>>>order to optionally generate the classes with caching, adding then this
>>
>>new
>>
>>>version in your repository?
>>>
>>>
>>>
>>>Sincerely,
>>>
>>>Massimo Pasina
>>>
>>>
>>>
>>>PD:I have just subscribed to [email protected]
>>>
>>
>>
>>
>>-------------------------------------------------
>>If you wish to unsubscribe from this list, please
>>send an empty message to the following address:
>>
>>[EMAIL PROTECTED]
>>-------------------------------------------------
>>
>
>
--
Syscon Ingenieurbüro für
Meß- und Datentechnik GmbH
Ralf Joachim
Raiffeisenstraße 11
D-72127 Kusterdingen
Germany
Tel. +49 7071 3690 52
Mobil: +49 173 9630135
Fax +49 7071 3690 98
Email: [EMAIL PROTECTED]
Web: www.syscon-world.de
-------------------------------------------------
If you wish to unsubscribe from this list, please
send an empty message to the following address:
[EMAIL PROTECTED]
-------------------------------------------------
