We own John Sisson thanks for the patch ... not me. I tested the patch
prior to the commit but it was John that created the patch just before
leaving on vacation. So, thank you John! I for one, am very grateful
to be able to build on windows ago!
Joe
Dain Sundstrom wrote:
Ever since I applied Prasad's patch for jaring the WEB-INF/classes of
our wars the server has been well under the windows path length limit.
Unfortunately, you couldn't build the server on windows since we
actually assemble the server in a deeply nested path.
I just applied a patch from Joe Bohn that shortened the names of nested
modules in ears. This has brought the longest path in our build to:
229 assemblies/j2ee-installer/target/geronimo-1.1-SNAPSHOT/repository/
geronimo/jsp-examples-tomcat/1.1-SNAPSHOT/jsp-examples-tomcat-1.1-
SNAPSHOT.car/WEB-INF/classes/org/apache/jsp/jsp2/jspattribute/
shuffle_jsp$shuffle_jspHelper.class
So as long as you checkout path is less then 27 characters you should
be able to build on windows.
-dain
On Apr 14, 2006, at 6:03 PM, John Sisson wrote:
It would be good if we could use precompiled jsps so the server is
always responsive - that first time hit often happens when you are
running a demo :-)
John
Prasad Kashyap wrote:
I could not use the ant task to precompile jsps. Using the following,
simply nothing happens.
<ant:taskdef classname="org.apache.jasper.JspC" name="ant:jasper2"
classpathref="jspc.classpath"/>
<ant:jasper2 validateXml="false" uriroot="${webroot}"
webXmlFragment="${webroot}/generated_web.xml" addWebXmlMappings="true"
outputDir="${outDir}" />
So by passing a few more args to our existing <ant:java > execution, I
was able to generate the xml file too. Using the option
addWebXmlMappings, I tried to merge the web.xmls but JspC cried that
it was a unrecognized option.
<!-- arg value="-addWebXmlMappings"/>
<arg value="true"/ -->
Looking at the JspC code,
http://svn.apache.org/repos/asf/tomcat/jasper/branches/tc5.0.x/
jasper2/src/share/org/apache/jasper/JspC.java
it seems like you can't pass addWebXmlMappings in a java command line
argument. It is not recognized or set by the setArgs(). Seems like a
jasper bug ?
So I added another goal to our deployment plugin called "mergeWebXML".
This goal merges the generated web.xml fragment with the existing one
from source.
When I spoke to the Tomcat guys, they feel that precompiling JSPs
doesn't buy us anything anymore, except for that first time hit. The
request mapper is supposedly fixed in 5.5.x versions and thus we
shouldn't see any other performance hit.
Anyways, I have uploaded the patch to
http://issues.apache.org/jira/browse/GERONIMO-1844
Cheers
Prasad
On 4/14/06, Prasad Kashyap <[EMAIL PROTECTED]> wrote:
OK. I'll take care of the JSP pages. I am done with jar'ing the
generated class files. I shall look into generating a web.xml and
merging it.
Cheers
Prasad
On 4/13/06, Dain Sundstrom <[EMAIL PROTECTED]> wrote:
Prasad got the patch working on all web applications. The next big
offenders left are the auto generated WebServices wrapper class and
the precompiled JSP pages. David Jencks has a patch for the
WebServices class, but we are waiting to get the TCK running so we
can judge the impact. I disabled precompiled JSP pages, as they
didn't work anyway. We generate the classes and compile them, but we
never added them to the class path and we are not merging generated-
web.xml into our web.xml. I'm hoping Prasad has time to tackle this
one.
With these patches in, except for the WS class, the longest paths
including server name is 224 and 217 characters:
geronimo-1.1-SNAPSHOT/repository/geronimo/daytrader-derby-jetty-
streamer-client/1.1-SNAPSHOT/daytrader-derby-jetty-streamer-
client-1.1-SNAPSHOT.car/activemq/activemq-ra/3.2.4-SNAPSHOT/rar/
activemq-optional-3.2.4-SNAPSHOT.jar
geronimo-1.1-SNAPSHOT/repository/geronimo/webconsole-jetty/1.1-
SNAPSHOT/webconsole-jetty-1.1-SNAPSHOT.car/geronimo-console-
framework-1.1-SNAPSHOT.war/WEB-INF/config/services/
PortletDefinitionRegistryService.properties
These will become 179 and 189 characters when we drop -SNAPSHOT from
all of the version numbers. I'm also hoping we can change the name
daytrader-derby-jetty-streamer-client to something shorter.
This is a lot better and will allow for at least 50 characters of
head room, which I hope is plenty for windows users.
As for next steps, I'd like to get the hot deploy service working
better. With the addition of the in-place deployment feature from
Gianny and the timestamp I added to the configuration, we should be
able to write a much better service. Once we have the better hot
deploy service, we will be able to implement the flat deployment
model that Hernan and others have suggested.
-dain
--
Joe Bohn
joe.bohn at earthlink.net
"He is no fool who gives what he cannot keep, to gain what he cannot
lose." -- Jim Elliot
