socket-developers Mailing List for SOCKET (Page 2)

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Two interesting links:

http://www.alt.ac.uk/altc2006/timetable/abstract.php?abstract_id=771

http://java.sun.com/developer/technicalArticles/J2SE/monitoring/

Brian

Aggie,
According to below:
http://developer.apple.com/documentation/Java/Conceptual/JavaPropVMInfoRef/
Articles/JavaVirtualMachineOptions.html
The max heap size default is 64M, 128 with the -server flag.
For tomcat the options are set in catalina.sh (CATALINA_OPTS) - this is a 
scary file.  We seem to be on defaults.
Regards,
Brian
Heap Size
-Xmssize in bytes 
    Sets the initial size of the Java heap. The default size is 2097152 (2MB). 
    The values must be a multiple of, and greater than, 1024 bytes (1KB). 
    (The -server flag increases the default size to 32M.)
-Xmnsize in bytes 
    Sets the initial Java heap size for the Eden generation. The default value 
    is 640K. (The -server flag increases the default size to 2M.)
-Xmxsize in bytes 
    Sets the maximum size to which the Java heap can grow. The default 
    size is 64M. (The -server flag increases the default size to 128M.) The 
    maximum heap limit is about 2 GB (2048MB).

Garbage Collection: Memory Usage

Note: Many of the garbage collection flags are dependent on the settings for 
the heap size. Make sure that you have the appropriate sizes set for the heap 
before fine-tuning how garbage collection uses that memory space.
-XX:MinHeapFreeRatio=percentage as a whole number 
    Modifies the minimum percentage of the heap kept free after garbage 
    collection. The default value of 40. If at least 40% of the heap is not 
    freed after garbage collection, the heap size increases.
-XX:MaxHeapFreeRatio=percentage as a whole number 
    Changes the maximum percentage of heap kept free after garbage 
    collection before the heap is shrunk. The default value is 70. This means 
    that if a garbage collection results in more than 70% of the heap being 
    freed, the heap size decreases.
-XX:NewSize=size in bytes 
    Sets the default size for the Eden generation of allocated objects. The 
    default value is 640K. (The -server flag increases the default size to 
    2M.)
-XX:MaxNewSize=size in bytes 
    Allows you to change the upper limit of the young object space in which 
    new objects are allocated. The default value is 640K. (The -server 
    flag increases the default size to 2M.)
-XX:NewRatio=value 
    Changes the ratio of new to old space sizes from the default value of 8 
    where the new space is 1/8 the size of the old space. 
-XX:SurvivorRatio=number 
    Modifies the ratio of size of the Eden to the survivor space from the 
    default of 10 where Eden is survivorRatio+2 times larger than the 
    survivor space.
-XX:TargetSurvivorRatio=percentage 
    Desired percentage of survivor space used after scavenge. Default is 50.
-XX:MaxPermSize=size in MB 
    Modifies the size of the permanent generation. The default is value is 32 
    (32MB).
-XX:-CleanPagesOnUncommit 
    Normally, when the garbage collection mechanism determines that the 
    heap can be shrunk, i.e., a lot of memory that had been used is no 
    longer needed by the program, the Java VM marks the memory as 
    clean. The operating system is then able to reclaim pages for use by 
    other processes without needing to page out the old memory content to 
    disk. This flag turns off this behavior so that these pages are not marked 
    as clean.

> What is the memory set at for the JVM?
> 
> Aggie
> 
> -----Original Message-----
> From: soc...@li...
> [mailto:soc...@li...] On Behalf Of Brian
> Peter Clark
> Sent: 10 August 2006 18:07
> To: soc...@li...
> Subject: [Socket-developers] PermGen error
> 
> Sockanalia,
> 
> I've just had to re-start tomcat on socket1 because of the following 
> error:
> 
> java.lang.OutOfMemoryError: PermGen space
> 
> I've mentioned somewhere that I've seen it before, all after heavy 
> war loading and unloading activity.
> 
> The heap is filling up and not draining out.  It suppose that it could 
> be that the garbage collector isn't quick enough to deal with the 
> legal class detritus.  Or there might be a memory leak.
> 
> It's something to keep an eye on.  Maybe try to put a tool on the 
> job to monitor the heap.
> 
> We could also increase the heap size on the JVM.  After all, we 
> have got a little spare capacity on socket1.
> 
> Guanxi seems happy enough now.
> 
> Regards,
> 
> Brian
> 
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job
> easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Socket-developers mailing list
> Soc...@li...
> https://lists.sourceforge.net/lists/listinfo/socket-developers
> 
> 
> -------------------------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> Socket-developers mailing list
> Soc...@li...
> https://lists.sourceforge.net/lists/listinfo/socket-developers
> 

What is the memory set at for the JVM?

Aggie

-----Original Message-----
From: soc...@li...
[mailto:soc...@li...] On Behalf Of Brian
Peter Clark
Sent: 10 August 2006 18:07
To: soc...@li...
Subject: [Socket-developers] PermGen error

Sockanalia,

I've just had to re-start tomcat on socket1 because of the following 
error:

java.lang.OutOfMemoryError: PermGen space

I've mentioned somewhere that I've seen it before, all after heavy 
war loading and unloading activity.

The heap is filling up and not draining out.  It suppose that it could 
be that the garbage collector isn't quick enough to deal with the 
legal class detritus.  Or there might be a memory leak.

It's something to keep an eye on.  Maybe try to put a tool on the 
job to monitor the heap.

We could also increase the heap size on the JVM.  After all, we 
have got a little spare capacity on socket1.

Guanxi seems happy enough now.

Regards,

Brian

-------------------------------------------------------------------------
Using Tomcat but need to do more? Need to support web services, security?
Get stuff done quickly with pre-integrated technology to make your job
easier
Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
_______________________________________________
Socket-developers mailing list
Soc...@li...
https://lists.sourceforge.net/lists/listinfo/socket-developers

Sockanalia,

I've just had to re-start tomcat on socket1 because of the following 
error:

java.lang.OutOfMemoryError: PermGen space

I've mentioned somewhere that I've seen it before, all after heavy 
war loading and unloading activity.

The heap is filling up and not draining out.  It suppose that it could 
be that the garbage collector isn't quick enough to deal with the 
legal class detritus.  Or there might be a memory leak.

It's something to keep an eye on.  Maybe try to put a tool on the 
job to monitor the heap.

We could also increase the heap size on the JVM.  After all, we 
have got a little spare capacity on socket1.

Guanxi seems happy enough now.

Regards,

Brian

I am going to do 0.9 development release on Thursday.
CVS Head will be tagged on Thursday the 10th August 2006 2pm

If anyone needs to include anything in 0.9 then commit before Thursday 2pm.

Ta
Atif.

It seems we don't have lurkers on our list.

Current people subscribed to this list are:
* atif
* rob
* brian
* aggie

Ta
Atif.

I have restructured the stylesheet directories on sourceforge.  

Under xslt there are now two directories: custom and generic.

Generic contains everything that was formerly under xslt.

At the same level as generic is the custom directory.  This is for 
stylesheets that have been customised for a particular service.  
Presently there are two directories under this: element-data and 
ssrun.  Underneath each of these is an xhtml directory holding the 
xslt stylesheets.

I have uploaded all the stylesheets sent to me by Rob yesterday 
as being the latest versions.

Please destroy all stylesheets that may be on your hard drive and 
synchronize to sourceforge.

These are, however, not the latest versions.  The true = 1 false = 2 
assignments for the radio buttons are still present. 

Can you fix these as soon as possible, Rob, and anything else that 
might not be current.  I would like to demo today, please.

Also, until a solution is found to the calendar javascript (CDATA) 
problem (any progress here?), the absolute url is being hardwired 
into the js code.  The hardwire is pointing to webforwireless on at 
least some of the stylesheets.  These should now point to 
socket.leeds.ac.uk.

I will now remove all the schemas from the socket-view directories 
so that the only copies of the schemas will be in socket-schemas.

socket-schema.jar contains the schemas if the view needs a 
reference at a future date.

Atif informs me that these are the latest versions.

Regards,

Brian

Syllabubs,

When creating customised stylesheets for any service we need to 
have a systematic naming convention so that we don't get 
confused in a sea of operation-lists.xsl.

The generics are

operation-list.xml
service-input.xsl
service-output.xsl

Any non-generics ought to be label with an abbreviation labelling 
the service concerned.

For ssrun, for example, we'll have:

ssrun-operation-list.xsl
ssrun-service-input.xsl
ssrun-service-output.xsl

The exception style sheets shouldn't need to be changed.

If customisation is carried out at the operation level then the 
operation ID should be appended to the name of the stylesheet as 
follows:

ssrun-operation-list_1.xsl
ssrun-service-inputt_1.xsl
ssrun-service-output_1.xsl

for operationID = 1, and so on.

I'll change the three main chemical elements style sheets
to

pt-operation-list.xsl
pt-service-input.xsl
pt-service-output.xsl

Same with the css style sheets:

ssrun-socket.css

if necessary, and

ssrun-socket_1.css

if necessary.

Regards,

Brian

-----Original Message-----
From: Brian P. Clark [mailto:b.p...@le...] On Behalf Of
bri...@bt...
Sent: 28 July 2006 09:14
To: 'soc...@li...'
Subject: countdown

Sennapods,

In the countdown to the kiss and tell, here's a basic starter for ten.

Today.

Rob: urls on the buttons to operation_list.html, operation_input.html
and operation_execute.html.  Plus when the response comes in it doesn't
seem aligned with the items above it in the table. Is there some
whitespace being inserted here?

Perhaps the exception stylesheets could be tarted up a bit.

Is the background gif off for aesthetic reasons?

How long do you think it might take to plug in the SVG graphical XSLT to
our output?  The bar chart, for example?

Atif: the engine.

Aggie & I:   Assis demo.  I'll try to build and set up the assis
producer on the dell today.  I'll also send a new socket jar that has
links to Atif's little test service and my ChemicalElementService to get
bods in action with real services.

I'll also add to the chemical elements service to test array output.

Regards,

Brian

Still having problems with email.

Chapatees, (Rob, mainly)

Could we change arg0 arg1 etc to Enter data:

Also we need to identify the data type associated with each input box.
This could either go something like (integer) directly underneath the
text box or Enter data (integer):

Whichever is the easier for the present.

Ultimately, we'll be able to assign custom values to all the strings in
the style sheet by JAXP injection of parameters but this won't be by
next Thursday.

Something for Atif to think about is taking the parameter names from the
Axis output, but I'm not sure that this would be worth the candle - a
lot of it will be String_0 and so on for Java to WSDL.   If we can
inject then this solves these problems..

Regards,

Brian

Sackbutters,

This gets more fun by the minute.  Although the periodic table is a 
"string in - string out" web service, because of the html output, we 
can serve out the response as a fragment of html markup.  This 
could be feedback for a wrong answer or a hyperlink or an image or 
....

Regards,

Brian

Roll up, roll up!

Test your knowledge of the symbols of the Periodic Table of the 
Elements:

http://195.10.244.166:8080/ptconsumer/consumer/operation_list.ht
ml

You generate an exception if the symbol doesn't exist.

Regards,

Brian

Seconals,

I think that we have poked our heads out, eyes blinking in the 
bright workaday sun.

The reason why the final round trip wasn't connecting, I'm pretty 
sure, is that the filter uses a (buffered) ByteArrayOutputStream 
from which it takes the data (as a byte array).

This works fine with posts & gets followed by dispatches.  
However, when Atif reconstructs the service response on the fly he 
injects this into the response as a string with a print method.  
Hence the filter sees nothing.

This only happens on the response stage and when a consumer 
exception is created on the fly.  The same "empty" byte array 
object is seen in both cases.

The easiest solution is for Atif to convert the response into a byte 
stream.

Sorted.

Regards,

Brian

ps I'm glad that this has come to me so early. I had said to Atif 
that I would not leave the office until I had spotted the critter.  I can 
now mark a 46 page essay that's lying on my desk...

Or I could go home for a large, stone-cold dry martini...

..............nnneeewowww!!

am in
  ----- Original Message -----=20
  From: bri...@bt...=20
  To: soc...@li...=20
  Sent: Tuesday, July 25, 2006 9:02 AM
  Subject: [Socket-developers] FW: Toolkits and Demonstrators Show and =
Tell,Manchester - 3rd August

  15 mins!

  =20

  -----Original Message-----
  From: JISC e-framework toolkit and demonstrator projects =
[mailto:JIS...@JI...] On Behalf Of Heather =
Williamson
  Sent: 25 July 2006 08:31
  To: JIS...@JI...
  Subject: Toolkits and Demonstrators Show and Tell, Manchester - 3rd =
August
  Importance: High

  =20

  Dear All

  =20

  The online booking form for the Toolkits and Demonstrators Show and =
Tell is now available at =
http://www.jisc.ac.uk/index.cfm?name=3Dform&formid=3D372637328 and the =
agenda for the day is attached to this email.=20

  Please register your attendance as soon as possible to ensure we can =
keep the venue fully informed of numbers.

  =20

  Further details about the meeting and what is expected from each =
project can be found below:

  =20

  =20

  TOOLKITS AND DEMONSTRATORS SHOW AND TELL MEETING

  =20

  Venue: Manchester Conference Centre (location details can be found on =
the booking form)

  Date: 3rd August 2006

  Accommodation: A limited number of rooms have been booked for the =
night of Wednesday 2nd August. If you require a room on this night, =
please ensure that you indicate this in the space provided on the =
booking form.

  =20

  AUDIENCE

  The toolkits and demonstrators show-and-tell meeting is a public event =
for projects to share their work.

  =20

  TIMINGS

  As there are 12 projects to get through, time will be strictly limited =
and project presentations must be no more than 15 minutes each. We would =
like to ask participants to keep issues and discussion points for the =
discussion session at the end of the day rather than eat into the time =
allocated for presentations.

  =20

  PRESENTATIONS

  All project presentations should be entirely made up of live =
demonstrations of working software. If for some reason that is not =
possible, projects may alternatively present a brief summary of their =
progress to-date and issues encountered using a 2 slide PowerPoint (or =
equivalent). The timing for your project presentation can be found on =
the attached agenda.

  =20

  PROJECT DISSEMINATION

  An information pack will be provided on the day which will include a =
brief summary of all the toolkit and demonstrator projects presenting on =
the day, or funded as part of this current stream of funding. Projects =
are encouraged to submit electronic versions of their slides, blog post, =
posters or demo outline by the end of this week so that they can be =
included in the pack (maximum of 2 sides of A4 please). If you wish to =
bring further information about your project with you on the day, there =
will also be a space for project leaflets/further information on a =
display table at the event which you can use.

  =20

  I look forward to seeing you all on the 3rd.

  =20

  Kind regards

  =20

  Heather

  =20

  =20

  ----------------------------------
  Heather Williamson
  JISC Programme Manager

  JISC Executive
  University of Bristol
  Tel: 07810 814468
  Email: h.w...@ji...=20

  =20

-------------------------------------------------------------------------=
-----

  =
-------------------------------------------------------------------------=

  Take Surveys. Earn Cash. Influence the Future of IT
  Join SourceForge.net's Techsay panel and you'll get the chance to =
share your
  opinions on IT & business topics through brief surveys -- and earn =
cash
  =
http://www.techsay.com/default.php?page=3Djoin.php&p=3Dsourceforge&CID=3D=
DEVDEV

-------------------------------------------------------------------------=
-----

  _______________________________________________
  Socket-developers mailing list
  Soc...@li...
  https://lists.sourceforge.net/lists/listinfo/socket-developers

Sequoias,

Java.lang.OutOfMemoryError: PermGen space

Just for future reference - a Java error I got after loading and unloading 
about 20 consumer.wars on the trot.

http://wiki.caucho.com/Java.lang.OutOfMemoryError:_PermGen_space

Brian

Secateurs,

The caching issue is now behind us.  Again, it is amazing what a 
little research will yield.

http://developer.mozilla.org/en/docs/Using_Firefox_1.5_caching#Sa
mple_code

For http traffic, Firefox wants Cache-Control: no-store instead of no-
cache. It uses no-cache for https.  IE Explorer seems happy with 
both.

I switched both Cache-Control and Pragma to no-store instead of 
no-cache and both browsers now behave almost impeccably and 
identically.

I'm off for a stiff whisky.

Regards,

Brian

Secateurs,

Not one of my more auspicious days!

The ServiceInput.xsl is now getting the operationID.  It is 
embarrassing to note that for around 2-3 hours the java has been 
trying to pass values to operationID2 when it had been changed 
back to operationID.

On a positive note, I have a more intimate relationship with the filter 
code.

In the end, a much ado about nothing much.... a long much ado...

So, Rob, you can now focus on singling out the one operation.

You'll need to post the operationID in the form.

When you've sorted out the operationID, you might spend half an 
hour putting the input data type under the input box like (integer), 
(boolean) and so on.  Seems like a reasonable default behaviour.

Atif has run the ASSIS wsdls through the system and two of them 
will present no problems.

There is test data provided with the ASSIS code -  we'll try to set 
up the provider up at the end of this week.

Onwards!

Regards,

Brian

Secateurs,

It appears that servlet 2.4 can actually map a filter on to a dispatch
(I had it in my mind that this was 2.5):

<filter-mapping>
<filter-name>Logging Filter</filter-name>
<url-pattern>/products/*</url-pattern>
<dispatcher>FORWARD</dispatcher>
<dispatcher>REQUEST</dispatcher>
</filter-mapping>
would result in the Logging Filter being invoked by client requests
starting /
products/... and underneath a request dispatcher forward() call where
the
request dispatcher has path commencing /products/....

With socket, by omitting the <dispatcher>REQUEST</dispatcher>, this
would eliminate the filter being invoked 'on the way in' which we don't
really need.

<filter-mapping>
<filter-name>ViewFilter</filter-name>
<url-pattern>/*</url-pattern>
<dispatcher>FORWARD</dispatcher>
</filter-mapping>

Regards,

Brian

I've edited this to introduce an outputValues element to match the 
outputValuesType and removed the inputBoolean etc.

We decided to introduce this because the output schema doesn't 
need the variable type information and to take arrays into account - 
outputValues is maxOccurs unbounded.

Regards,

Brian

<?xml version="1.0" encoding="UTF-8"?>
<xs:schema 
xmlns:guid="http://socketelf.org/guid/output" 
xmlns:xs="http://www.w3.org/2001/XMLSchema" 
targetNamespace="http://socketelf.org/guid/output
" elementFormDefault="qualified">
	<xs:element name="ServiceOutput" 
type="guid:ServiceOutputType">
		<xs:annotation>
			<xs:documentation>SOCKET schema 
describing data from service response. 
</xs:documentation>
		</xs:annotation>
	</xs:element>
	<xs:complexType name="ServiceOutputType">
		<xs:sequence>
			<xs:element name="serviceName" 
type="xs:string"/>
			<xs:element name="operation" 
type="guid:operationType"/>
		</xs:sequence>
	</xs:complexType>
	<xs:complexType name="operationType">
		<xs:sequence>
			<xs:element name="operationName" 
type="xs:string"/>
			<xs:element name="operationID" 
type="xs:int"/>
			<xs:element name="outputElements" 
type="guid:outputElementsType" />
		</xs:sequence>
	</xs:complexType>
	<xs:complexType name="outputElementsType">
		<xs:sequence>
			<xs:element name="outputElement" 
type="guid:outputElementType" minOccurs="1" 
maxOccurs="unbounded"/>
		</xs:sequence>
	</xs:complexType>
	<xs:complexType name="outputElementType">
		<xs:sequence>
			<xs:element name="outputValues" 
type="guid:outputValuesType" nillable="true"/>
			<xs:element name="label" 
type="xs:string" nillable="true"/>
			<xs:element name="description" 
type="xs:string" nillable="true"/>
		</xs:sequence>
	</xs:complexType>
	<xs:complexType name="outputValuesType">
		<xs:sequence>
			<xs:element name="value" 
type="xs:string" minOccurs="1" 
maxOccurs="unbounded"/>
		</xs:sequence>
	</xs:complexType>
</xs:schema>

Secateurs,

Following some discussions between myself and Aggie, the 
following are the targets for our demonstration at the Kiss and Tell 
on August 4.

1.  SOCKET Questionmark Perception in Bodington via QMShibb.

2.  4 SOCKET services related to a JISC assessment project 
(ASSIS).  I attach three of the wsdls, the fourth client will be formed 
from the published client java classes produced by a wsdl2java run.

3.  I'd also like to demo the JISC Interloc chat application as an 
application-wrapped SOCKET client talking to a Jabber server.

The heart of the show will be a demo of the complete WSDL to 
SOCKET service using a simple example - maybe the Periodic 
Table service.

I'd also like to show how we can alter the labelling of the input 
forms through an XSLT servlet filter.

I think that'll fill up the 30 minutes.

Regards,

Brian

Good stuff about how to get axis  to handle a service that returns an
array of beans.

http://www.mail-archive.com/axi...@xm.../msg07711.html

Regards,

Brian

A new world record!

203 hits and only 18.58 hrs (Atif says spiders have been busy).

Brian

Atif,

I think we ought to give some serious consideration to putting a wsdd
file into rose, even if it just sits there and we configure it manually.
Or have a wsdd bean with some setters and getters that the console might
call.  I don't think it would mean anything much in terms of changes to
the present codebase and it would give us the flexibility of, for
example, invoking the SimpleSessionHandler.  This would give us access
to a whole lot more services.

Something for July perhaps.

Regards,

Brian

Rob,

On the operation response page, you've already got a button that takes
the user back to the list of all operations.  Could you please put
another button that goes back to the input page of the same operation.
A user will often want to submit the same operation multiple times with
different parameters.  This button will make life a lot easier.

I've seen some bioinformatics programs that give you one entry from a
result set each time you submit an operation (SimpleSessionHandler
invoked on client and server flows).

Cheers,

Brian

Secateurs,

I think we'll need one more stylesheet - for soap faults and
socketprojectexception reports.  The client ought not to see the naked
exception, but we'll have to produce a nice screen to say that something
has gone wrong, and that it's been logged and that a nice person is
looking into it.  There are a number of possible faults - network down,
service producer fail, security exception (not handling the session
properly), or invalid soap or message response.  Perhaps some extensions
to SocketProjectException are required?

Atif will add another two elements to the response-type enumeration -
soap fault and socketprojectexception - and I'll knock up a simple
schema for the xml on which Rob's stylesheet can operate. 

In the service-oriented world a delegate class is often used here to
buffer the client from the raw exception.  This would normally sit on
the client side.  In this case we might have a servlet filter as the
delegate. It intercepts any fault/error messages and renders an
appropriate response.

Regards,

Brian

socket-developers Mailing List for SOCKET (Page 2)

socket-developers — Mailing list for SOCKET developers