Re: Migration from 5.5.20 to 6.0.10: parser issue on application deployment

2007-03-04 Thread Etienne Giraudy

To see how "standard" and "legal" this usage is, you can try enabling
the security manager (the only way to control writing to system
properties - which are always JVM wide, so since Tomcat uses JAXP,
Tomcat cannot avoid being affected to some extent when a webapp
changes the parser factory - is to use the security manager) :) The
other problem is that this by defeinition won't behave the same in
different VMs (1.4, and same if you have a 5.0 VM with the
compatibility pack for TC 5.5).


Not allowing the application to modify this system property is not an
option for me.



You can send me a WAR to test however, I would be interested to look
at this alleged difference in behavior.



I've filled a bug report to which I attached a simple web app for
reproducing the issue:
http://issues.apache.org/bugzilla/show_bug.cgi?id=41759


Etienne

-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Re: Migration from 5.5.20 to 6.0.10: parser issue on application deployment

2007-03-04 Thread Rémy Maucherat

On 3/4/07, Etienne Giraudy <[EMAIL PROTECTED]> wrote:

I guess that the point that is questionnable here is the way the API
is designed: modifying the system property 'legal' and, AFAIK, it is
the only way to choose the parser implementation we want to use
(http://java.sun.com/j2se/1.5.0/docs/api/javax/xml/parsers/SAXParser.html).

From that point of view, shouldn't Tomcat protect itself against
bad-designed standard APIs usage?


To see how "standard" and "legal" this usage is, you can try enabling
the security manager (the only way to control writing to system
properties - which are always JVM wide, so since Tomcat uses JAXP,
Tomcat cannot avoid being affected to some extent when a webapp
changes the parser factory - is to use the security manager) :) The
other problem is that this by defeinition won't behave the same in
different VMs (1.4, and same if you have a 5.0 VM with the
compatibility pack for TC 5.5).

You can send me a WAR to test however, I would be interested to look
at this alleged difference in behavior.

Rémy

-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Re: Migration from 5.5.20 to 6.0.10: parser issue on application deployment

2007-03-04 Thread Rémy Maucherat

On 3/3/07, Etienne Giraudy <[EMAIL PROTECTED]> wrote:

Shall this be considered as a regression as in that case tomcat
configuration is somehow altered by a web app?
(in that case I'll fill a bug in bugzilla))


I don't think there can be a change of behavior in this sort of thing
between TC 5.5 and 6.0.

Rémy

-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



Re: Migration from 5.5.20 to 6.0.10: parser issue on application deployment

2007-03-03 Thread Etienne Giraudy

On 3/3/07, Caldarale, Charles R <[EMAIL PROTECTED]> wrote:

> From: Etienne Giraudy [mailto:[EMAIL PROTECTED]
> Subject: Migration from 5.5.20 to 6.0.10: parser issue on
> application deployment
>
> One of the web app running on that server includes
> xercesImpl.jar and use it through modifying the system
> property javax.xml.parsers.SAXParserFactory.

That behavior looks rather questionable to me.  Having a webapp modify a
global property that has the potential of affecting everything running
in that JVM - including other webapps and the container - seems like a
very risky strategy, raising serious compatibility and operability
issues.  The fact that it happened to work in one version of one
container doesn't imply that it's an appropriate thing to do.


I guess that the point that is questionnable here is the way the API
is designed: modifying the system property 'legal' and, AFAIK, it is
the only way to choose the parser implementation we want to use
(http://java.sun.com/j2se/1.5.0/docs/api/javax/xml/parsers/SAXParser.html).


From that point of view, shouldn't Tomcat protect itself against

bad-designed standard APIs usage?

Etienne

-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]



RE: Migration from 5.5.20 to 6.0.10: parser issue on application deployment

2007-03-03 Thread Caldarale, Charles R
> From: Etienne Giraudy [mailto:[EMAIL PROTECTED] 
> Subject: Migration from 5.5.20 to 6.0.10: parser issue on 
> application deployment
> 
> One of the web app running on that server includes 
> xercesImpl.jar and use it through modifying the system
> property javax.xml.parsers.SAXParserFactory.

That behavior looks rather questionable to me.  Having a webapp modify a
global property that has the potential of affecting everything running
in that JVM - including other webapps and the container - seems like a
very risky strategy, raising serious compatibility and operability
issues.  The fact that it happened to work in one version of one
container doesn't imply that it's an appropriate thing to do.

 - Chuck


THIS COMMUNICATION MAY CONTAIN CONFIDENTIAL AND/OR OTHERWISE PROPRIETARY
MATERIAL and is thus for use only by the intended recipient. If you
received this in error, please contact the sender and delete the e-mail
and its attachments from all computers.

-
To start a new topic, e-mail: users@tomcat.apache.org
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]