You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@tomcat.apache.org by Mark Thomas <ma...@apache.org> on 2012/09/09 21:01:10 UTC
Resources - JarScanner / WebappClassLoader
This is issue d) in Konstantin's comments in TOMCAT-NEXT.txt
Konstantin is right to point out the inefficiencies in JarScanner. There
is no need to scan WEB-INF/lib and WEB-INF/classes as the class path
from the WebappClassLoader already includes them.
However, I think this can be considered separately from the Resources
implementation. My suggestion is to open a BZ enhancement request for this.
My one comment is that we'll need to be careful to retain the current
functionality of having the option to scan just WEB-INF or the entire
class path.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: Resources - JarScanner / WebappClassLoader
Posted by Mark Thomas <ma...@apache.org>.
On 10/09/2012 17:34, Mark Thomas wrote:
> On 10/09/2012 15:11, Mark Thomas wrote:
>> On 09/09/2012 20:01, Mark Thomas wrote:
>>> This is issue d) in Konstantin's comments in TOMCAT-NEXT.txt
>>>
>>> Konstantin is right to point out the inefficiencies in JarScanner. There
>>> is no need to scan WEB-INF/lib and WEB-INF/classes as the class path
>>> from the WebappClassLoader already includes them.
>>>
>>> However, I think this can be considered separately from the Resources
>>> implementation. My suggestion is to open a BZ enhancement request for this.
>>>
>>> My one comment is that we'll need to be careful to retain the current
>>> functionality of having the option to scan just WEB-INF or the entire
>>> class path.
>>
>> Time for plan B. Already :)
>>
>> The Virtual[Loader|DirContext] unit tests are proving very useful at
>> ferreting out edge cases. The JarScanner currently depends on a
>> consistent URL space under WEB-INF/classes which is no longer the case.
>>
>> Fixing the consistent URL space issue with the JarScanner is an ideal
>> opportunity to address the above issue as well. I'll patch trunk and
>> then port to 7.0.x and the sandbox.
>
> The edge cases mean the fix for trunk and trunk-resources is turning out
> to be very different. I'm going to focus on trunk-resources.
This is complete for trunk-resources.
I don't intend addressing it in trunk as a different solution would be
required. Neither do I intend addressing it in 7.0.x. On this basis, I
believe this issue is now resolved.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: Resources - JarScanner / WebappClassLoader
Posted by Mark Thomas <ma...@apache.org>.
On 10/09/2012 15:11, Mark Thomas wrote:
> On 09/09/2012 20:01, Mark Thomas wrote:
>> This is issue d) in Konstantin's comments in TOMCAT-NEXT.txt
>>
>> Konstantin is right to point out the inefficiencies in JarScanner. There
>> is no need to scan WEB-INF/lib and WEB-INF/classes as the class path
>> from the WebappClassLoader already includes them.
>>
>> However, I think this can be considered separately from the Resources
>> implementation. My suggestion is to open a BZ enhancement request for this.
>>
>> My one comment is that we'll need to be careful to retain the current
>> functionality of having the option to scan just WEB-INF or the entire
>> class path.
>
> Time for plan B. Already :)
>
> The Virtual[Loader|DirContext] unit tests are proving very useful at
> ferreting out edge cases. The JarScanner currently depends on a
> consistent URL space under WEB-INF/classes which is no longer the case.
>
> Fixing the consistent URL space issue with the JarScanner is an ideal
> opportunity to address the above issue as well. I'll patch trunk and
> then port to 7.0.x and the sandbox.
The edge cases mean the fix for trunk and trunk-resources is turning out
to be very different. I'm going to focus on trunk-resources.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org
Re: Resources - JarScanner / WebappClassLoader
Posted by Mark Thomas <ma...@apache.org>.
On 09/09/2012 20:01, Mark Thomas wrote:
> This is issue d) in Konstantin's comments in TOMCAT-NEXT.txt
>
> Konstantin is right to point out the inefficiencies in JarScanner. There
> is no need to scan WEB-INF/lib and WEB-INF/classes as the class path
> from the WebappClassLoader already includes them.
>
> However, I think this can be considered separately from the Resources
> implementation. My suggestion is to open a BZ enhancement request for this.
>
> My one comment is that we'll need to be careful to retain the current
> functionality of having the option to scan just WEB-INF or the entire
> class path.
Time for plan B. Already :)
The Virtual[Loader|DirContext] unit tests are proving very useful at
ferreting out edge cases. The JarScanner currently depends on a
consistent URL space under WEB-INF/classes which is no longer the case.
Fixing the consistent URL space issue with the JarScanner is an ideal
opportunity to address the above issue as well. I'll patch trunk and
then port to 7.0.x and the sandbox.
Mark
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@tomcat.apache.org
For additional commands, e-mail: dev-help@tomcat.apache.org