You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@ofbiz.apache.org by Adam Heath <do...@brainfood.com> on 2009/04/14 22:27:42 UTC
Re: svn commit: r757671 - in /ofbiz/trunk: ./ framework/common/servicedef/
framework/example/testdef/assertdata/ framework/service/src/org/ofbiz/service/
framework/service/src/org/ofbiz/service/jms/ framework/webapp/src/org/ofbiz/webapp/control/
framework/...
Adam Heath wrote:
> jonesde@apache.org wrote:
>> Author: jonesde
>> Date: Tue Mar 24 06:27:03 2009
>> New Revision: 757671
I've had a chance to look over this change, the removal of
WebAppDispatcher. And, it could break some code.
WebAppDispatcher has a check for a null classloader; if so, it uses
the Thread's classloader.
However, the standard GenericDispatcher does *not* do this; instead,
it uses the classloader of GenericDispatcher.class.
So, I consider this change to actually introduce a *regression*, even
if it doesn't actually cause any bugs in ofbiz trunk.
Additionally, ofbiz maintains a global cache of dispatchers by name.
This is a big nono for webslinger. Webslinger has a *single* lookup
point, and then everything else it uses is keyed off of that. This
allows for recycling to be done, by just removing the top-level reference.
Re: svn commit: r757671 - in /ofbiz/trunk: ./ framework/common/servicedef/ framework/example/testdef/assertdata/ framework/service/src/org/ofbiz/service/ framework/service/src/org/ofbiz/service/jms/ framework/webapp/src/org/ofbiz/webapp/control/ framework/
Posted by Jacques Le Roux <ja...@les7arts.com>.
Please Adam,
It would be cool if you could copy at least some lines of the previous msg in the thread. This is not specific to this msg of course
;o)
TIA
Jacques
From: "Adam Heath" <do...@brainfood.com>
> Adam Heath wrote:
>> jonesde@apache.org wrote:
>>> Author: jonesde
>>> Date: Tue Mar 24 06:27:03 2009
>>> New Revision: 757671
>
> I've had a chance to look over this change, the removal of
> WebAppDispatcher. And, it could break some code.
>
> WebAppDispatcher has a check for a null classloader; if so, it uses
> the Thread's classloader.
>
> However, the standard GenericDispatcher does *not* do this; instead,
> it uses the classloader of GenericDispatcher.class.
>
> So, I consider this change to actually introduce a *regression*, even
> if it doesn't actually cause any bugs in ofbiz trunk.
>
> Additionally, ofbiz maintains a global cache of dispatchers by name.
> This is a big nono for webslinger. Webslinger has a *single* lookup
> point, and then everything else it uses is keyed off of that. This
> allows for recycling to be done, by just removing the top-level reference.
>