thanks.
I'll provide backup if you need. and do regression test on my side after we
patch it.
2009/11/5 Hubert Iwaniuk <neotyk_at_kungfoo.pl>
> First look at classloaders produced by old and new approach shows following
> differences:
> war contents in old is referenced vis jar:file: protocol while with new one
> via file: protocol to exploded location.
> Same stands for WEB-INF/classes.
> So it should not produce any difference, but apparently it somehow does.
> I'll try looking into it a bit more.
>
> Cheers,
> Hubert
>
>
>
> On Thu, Nov 5, 2009 at 8:27 PM, Survivant 00 <survivant00_at_gmail.com>wrote:
>
>> I couldn't wait until tonight.
>>
>> I been able to make it works, but I had to change a little code. (now I'm
>> pretty sure that something missing in the ClassLoader).
>>
>> here what I changed : (call the older classloader instead for this test)
>>
>> public void deployWar(
>> String location, String context, URLClassLoader
>> serverLibLoader, WebApp defaultWebApp) throws DeployException {
>> String ctx = context;
>> if (ctx == null) {
>> ctx = getContext(location);
>> int i = ctx.lastIndexOf('.');
>> if (i > 0) {
>> ctx = ctx.substring(0, i);
>> }
>> }
>> Map.Entry<String, URLClassLoader> loaderEntry = null;
>> try {
>> loaderEntry = explodeAndCreateWebAppClassLoader(location,
>> serverLibLoader);
>> } catch(Exception e){
>> e.printStackTrace();
>> }
>>
>> WarDeploymentConfiguration config = new
>> WarDeploymentConfiguration(ctx, loaderEntry.getValue(), defaultWebApp);
>> if (logger.isLoggable(Level.FINER)) {
>> logger.log(Level.FINER, String.format("Configuration for
>> deployment: %s.", config));
>> }
>> deployer.deploy(ws, new File(location).toURI(), config);
>> }
>>
>>
>> and spring found the files.
>>
>>
>> Do you want me to go deeper ? (into the code :0 )
>>
>
>
--
Vous pouvez me suivre sur Twitter / You can follow me on Twitter :
http://twitter.com/survivant