tm4j-users — Forum for users of the TM4J API or apps developed by this project.

[TM4J-users] TM4Web: factory.props missing

[Jonas G. <gr...@gm...>]

Hi,

After quite a bit of trial and error I've finally got TM4Web/Velocity sort
of up and running. However, when I try to process a topic map i get the
following error output:

ant -f batch.xml -Dtopicmap=test.xtm topicmap
Buildfile: batch.xml

topicmap:
[copy] Copying 1 file to C:\TEMP\tm4web\build\batch\temp
[java] 0 [main] WARN org.tm4j.vtl.factory.BaseFactory - Could
not locate properties file: factory.props
[java] Cannot find CatalogManager.properties
[java] Processing explicit mergeMap directives...
[java] Merging referenced topic maps...
[java] All external maps merged!
[java] Making merged map consistent...
[java] java.lang.StackOverflowError
[java] Exception in thread "main"

BUILD SUCCESSFUL
Total time: 3 seconds

Did I misunderstand something here? I've searched for a file named
factory.props which, as anticipated, is not to be found anywhere. All
references and paths to various jars should be in place.

I should add that I'm running the binary package including the Velocity
files (installed from file tm4web-velocity-bin-0.2rc1.zip) with version
0.9.6 (binary) of TM4J, that I have installed Apache Ant separately, and
that I'm running Apache Tomcat as a part of the Ontopia Omnigator.

I really hope I'm making some kind of obvious mistake..

- jonas.

RE: [TM4J-users] topic map sizes and backends

[Richard G. <ric...@ya...>]

Thanks, Conal.

 --- Conal Tuohy <Con...@vu...> wrote: 
> Richard Gregory wrote:
> 
> > The largest I've created so far is 57MB as an XTM file (29347 
> > topics, 73756 associations and 26905
> > occurrences). I'm working exclusively with XTM files and the 
> > in memory backend - I long since gave
> > up on hibernate and ozone for these sort of size topic maps.
> 
> I've been using the In-Memory back-end with a topic map which serializes to a 27MB XTM file with
> 13k topics, 31k associations, 62k roleSpecs, and 9k occurrences. It seems to consume about 270MB
> in memory.
> 
> 
> -------------------------------------------------------
> This SF.net email is sponsored by: IT Product Guide on ITManagersJournal
> Use IT products in your business? Tell us what you think of them. Give us
> Your Opinions, Get Free ThinkGeek Gift Certificates! Click to find out more
> http://productguide.itmanagersjournal.com/guidepromo.tmpl
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
>  


	
	
		
___________________________________________________________ALL-NEW Yahoo! Messenger - all new features - even more fun!  http://uk.messenger.yahoo.com

RE: [TM4J-users] topic map sizes and backends

[Conal T. <Con...@vu...>]

Richard Gregory wrote:

> The largest I've created so far is 57MB as an XTM file (29347=20
> topics, 73756 associations and 26905
> occurrences). I'm working exclusively with XTM files and the=20
> in memory backend - I long since gave
> up on hibernate and ozone for these sort of size topic maps.

I've been using the In-Memory back-end with a topic map which serializes =
to a 27MB XTM file with 13k topics, 31k associations, 62k roleSpecs, and =
9k occurrences. It seems to consume about 270MB in memory.

[TM4J-users] topic map sizes and backends

[Richard G. <ric...@ya...>]

Hi All,

I know the performance of the various tm4j backends with large topic maps has been discussed
previously, but I'm interested in getting an idea of what the largest topic maps are that people
are using with the different backends. 

The largest I've created so far is 57MB as an XTM file (29347 topics, 73756 associations and 26905
occurrences). I'm working exclusively with XTM files and the in memory backend - I long since gave
up on hibernate and ozone for these sort of size topic maps.

I'm currently writing up my MSc dissertation on topic maps, and thought it may be usefull to be
able to include some information, by way to comparison, indicating the sort of topic map sizes
other people are using with the different backends, and getting acceptable performace for.

Thanks in advance,

Richard.



	
	
		
___________________________________________________________ALL-NEW Yahoo! Messenger - all new features - even more fun!  http://uk.messenger.yahoo.com

RE: [TM4J-users] TM4Web Template Configuration WAS RE: TM4Web/Velocity 0.2rc1 supports Hibernatebackends?

[Conal T. <co...@pa...>]

Kal:

> That sounds interesting - I haven't got really deeply into Cocoon so
> JXPath just completely passed me by :)

I've been using Cocoon for ages but I only really discovered JXPath (and
Cocoon's JXTemplateGenerator) recently myself. I think the JXTemplate
language  is very nice ... quite similar to XSLT, though without the
pattern-matching templates.

> In terms of associating templates with classes of topics for the
> Velocity integration, currently that is done by an external (XML)
> configuration file. The format allows you to specify a topic by its
> subject indicator and then specify a template. You can scope the
> application of the template to apply to:
>   1) Just the specified topic
>   2) Any instance of the topic
>   3) Any subclass of the topic
> (2) respects the subclass-superclass hierarchy, and the code
> attempts to
> locate the closest class in the hierarchy. So if you have a template
> person.vm scoped for instances of a topic 'Person' and then have 'Man'
> and 'Woman' as subclasses, person.vm would be used to render instances
> of 'Person', 'Man' and 'Woman'.

That's interesting. What we've implemented so far is just item (2) in the
list. I've also used scopes to scope the jxtemplate occurrences, so you can
request a topic to be rendered in one of a variety of scopes (something like
the @mode attribute in xslt templates I suppose). I haven't done a lot of
scopes yet ... it's mostly been for debugging. We've really just started the
user interface.

I've got a little controller written in javascript which looks for the
nearest jxtemplate occurrence of the topic's class (or classes). It struck
me when I wrote it in javascript that javascript was not the best language
for it. It wasn't huge, but it was 10 or 20 lines of code. I've been
thinking of something more like templates in XSLT, where each template would
have a jxpath expression to identify matching topics, and another jxpath
expression to match scopes. The controller would pick the first template
whose expressions matched the content and scope topics.

We'll definitely need something more than just (2) above ... amongst other
things, many of the topics in our topic maps represent parts of books or
magazines in a particular series, and we want to be able to use a visual
theme that matches the series.

> One thing that I have found really useful in the Velocity framework is
> the concept of the toolbox that contains utility classes that are
> initialised either per-app, per-session or per-request. It makes it
> quite easy to integrate helpers like the TologTool.

Hmmm .. so the toolbox is a container for global variables, sessions, local
variables, etc? Yes Cocoon does have equivalents I think.

Cocoon is big on separation of concerns and I think that "best practice" in
Cocoon is to do this kind of stuff prior to invoking the actual rendering
templates, as a separate pipeline stage with a transformer that marshalls
xml into a session, or else in javascript. The javascript implementation
uses continuations to invert the typical flow of control. You write ordinary
"procedural" javascript that "sends" the output of a Cocoon pipeline to the
browser (it can pass parameters, e.g. a content Topic and an array of scope
Topics to the pipeline). When it "sends" the user, the javascript code can
block to wait for user input, in which case the javascript is paused and
program state is stored as a continuation object. When the user clicks a
link on the client side, your continuation is revived and the javascript
code picks up again from the next line. It looks like the server sends and
the browser replies. Anyway ... with this technique these helper objects can
be just ordinary variables in the javascript code. But I'm not an expert on
this at all ... I've hardly used it actually.

> Also I find it
> useful to have a simple Factory helper (because there doesn't
> seem to be
> a way to initialise new objects in Velocity's templating language) -
> when you combine this with Mango's functional classes you can
> make lots
> of reusable helper objects that make writing the templates quite easy.
> Does the Cocoon framework have similar helpers ?

JXPath is extensible and there's a standard "constructor" helper function.
That sounds very similar:
http://jakarta.apache.org/commons/jxpath/apidocs/org/apache/commons/jxpath/f
unctions/ConstructorFunction.html
You can add functions to JXPath by implementing the interface "Function":
http://jakarta.apache.org/commons/jxpath/apidocs/org/apache/commons/jxpath/F
unction.html
And you can add iterators and operators and so on. I think on the whole it's
probably very similar to programming with Mango (not that I've every used
Mango, but I've read about it).

You can add axes to the JXPath graph in various ways: just by writing plain
java beans, or with standard Collection classes, DOMs, etc.

Cheers

Con

[TM4J-users] TM4Web Template Configuration WAS RE: TM4Web/Velocity 0.2rc1 supports Hibernatebackends?

[Kal A. <ka...@te...>]

On Tue, 2004-09-28 at 05:47, Conal Tuohy wrote:

> It's only vaguely related to Kal's post, but I thought I'd mention:
> 
> I haven't used the velocity templating engine, but I have used Cocoon to do templating. I've used Cocoon's JXTemplateGenerator, which is rather like XSLT with JXPath instead of XPath. JXPath gives you path-based access to Java object-graphs, e.g. x.getFoo().getBar() => "x/foo/bar". So this is similar to Velocity, and similarly I have templates associated with various topic map objects. 
> 
> I am keeping references to the individual JXTemplate files in the actual topic map, where they appear as occurrences of class topics. I had considered keeping a list of templates associated with different topics when I realised that this is what topic maps are for :-) I don't have a very sophisticated setup yet, but it does seem to me that referring to templates in the map itself is potentially a powerful technique. How does it work with the Velocity integration? Can you attach Velocity templates to particular classes of topics, for instance?
> 
> 

That sounds interesting - I haven't got really deeply into Cocoon so
JXPath just completely passed me by :)

In terms of associating templates with classes of topics for the
Velocity integration, currently that is done by an external (XML)
configuration file. The format allows you to specify a topic by its
subject indicator and then specify a template. You can scope the
application of the template to apply to:
  1) Just the specified topic
  2) Any instance of the topic
  3) Any subclass of the topic
(2) respects the subclass-superclass hierarchy, and the code attempts to
locate the closest class in the hierarchy. So if you have a template
person.vm scoped for instances of a topic 'Person' and then have 'Man'
and 'Woman' as subclasses, person.vm would be used to render instances
of 'Person', 'Man' and 'Woman'.

One thing that I have found really useful in the Velocity framework is
the concept of the toolbox that contains utility classes that are
initialised either per-app, per-session or per-request. It makes it
quite easy to integrate helpers like the TologTool. Also I find it
useful to have a simple Factory helper (because there doesn't seem to be
a way to initialise new objects in Velocity's templating language) -
when you combine this with Mango's functional classes you can make lots
of reusable helper objects that make writing the templates quite easy.
Does the Cocoon framework have similar helpers ?

Cheers,

Kal
-- 
Kal Ahmed <ka...@te...>
techquila

RE: [TM4J-users] TM4Web/Velocity 0.2rc1 supports Hibernatebackends?

[Conal T. <Con...@vu...>]

Kal Ahmed wrote:

> I've checked in some changes for the TM4Web/Velocity package which
> provides a basic level of support for listing the topic maps in a
> database. You need a configuration like this:
>=20
> <provider name=3D"hibernate"=20
> factory=3D"org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
>     <provider-properties>
>       <property name=3D"hibernate.connection.url"=20
> value=3D"jdbc:mysql://localhost/tm4mail"/>
>       <property name=3D"hibernate.connection.username" value=3D"***"/>
>       <property name=3D"hibernate.connection.password" value=3D"***"/>
>       <property name=3D"hibernate.dialect"=20
> value=3D"net.sf.hibernate.dialect.MySQLDialect"/>
>       <property name=3D"hibernate.connection.driver_class"=20
> value=3D"com.mysql.jdbc.Driver"/>
>     </provider-properties>  =20
>     <topicmaps template=3D"topicmap.vm"/>
>   </provider>
>=20
> Note the extra <topicmaps/> element that is required. The template
> attribute specifies the template to be used for the main=20
> topicmap page.
> With this update, you are forced to always use topic.vm to render
> topics, but I will work on adding support like you have in the normal
> <topicmap> element to specify the template(s) to use to render topics.

It's only vaguely related to Kal's post, but I thought I'd mention:

I haven't used the velocity templating engine, but I have used Cocoon to =
do templating. I've used Cocoon's JXTemplateGenerator, which is rather =
like XSLT with JXPath instead of XPath. JXPath gives you path-based =
access to Java object-graphs, e.g. x.getFoo().getBar() =3D> "x/foo/bar". =
So this is similar to Velocity, and similarly I have templates =
associated with various topic map objects.=20

I am keeping references to the individual JXTemplate files in the actual =
topic map, where they appear as occurrences of class topics. I had =
considered keeping a list of templates associated with different topics =
when I realised that this is what topic maps are for :-) I don't have a =
very sophisticated setup yet, but it does seem to me that referring to =
templates in the map itself is potentially a powerful technique. How =
does it work with the Velocity integration? Can you attach Velocity =
templates to particular classes of topics, for instance?

RE: [TM4J-users] exporting subject indicators

[Conal T. <Con...@vu...>]

Kal wrote:

> It definitely has to be a <subjectIndicatorRef> element. The=20
> <topicRef>
> element won't add the subjectIndicator property to your topic.=20
>=20
> My advice would be to make it absolutely explicit in your=20
> input pipeline
> that you are creating subject indicators and make full URIs:
>=20
> <topic id=3D"foo">
>   <subjectIdentity>
>     <subjectIndicatorRef
> =
xlink:href=3D"http://nztec.org/tm/harvest/xtm/nztec-ontology.xtm#foo"/>
>   </subjectIdentity>
> </topic>
>=20
> It may seem like a bit of overkill, but its the most robust way to
> ensure that what you mean is what is read by the XTM parser !

Cheers Kal. That's worked beautifully. Now I can save and reload my TM =
and still merge with it.

In fact I can specify the base locator when harvesting, so I just =
needed:

<topic id=3D"foo">
  <subjectIdentity>
    <subjectIndicatorRef xlink:href=3D"#foo"/>
  </subjectIdentity>
</topic>

Con

Re: [TM4J-users] TM4Web/Velocity 0.2rc1 supports Hibernate backends?

[Mathias R. <mat...@we...>]

Hello Kal,

thanks a lot - it works!

Regards,
Mathias

Kal Ahmed wrote:

>Hi Mathias,
>
>I've checked in some changes for the TM4Web/Velocity package which
>provides a basic level of support for listing the topic maps in a
>database. You need a configuration like this:
>
><provider name="hibernate" 
>factory="org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
>    <provider-properties>
>      <property name="hibernate.connection.url" 
>value="jdbc:mysql://localhost/tm4mail"/>
>      <property name="hibernate.connection.username" value="***"/>
>      <property name="hibernate.connection.password" value="***"/>
>      <property name="hibernate.dialect" 
>value="net.sf.hibernate.dialect.MySQLDialect"/>
>      <property name="hibernate.connection.driver_class" 
>value="com.mysql.jdbc.Driver"/>
>    </provider-properties>   
>    <topicmaps template="topicmap.vm"/>
>  </provider>
>
>Note the extra <topicmaps/> element that is required. The template
>attribute specifies the template to be used for the main topicmap page.
>With this update, you are forced to always use topic.vm to render
>topics, but I will work on adding support like you have in the normal
><topicmap> element to specify the template(s) to use to render topics.
>
>In tmbrowse, all topic maps are listed on the index page with their
>title as the full base URL of the topic map, so you end up with the URL
>listed twice (because by default the index page shows the source URL as
>well) - you may want to edit the listmaps.vm template to shorten this.
>
>I hope this gets you the basic functionality you need right now - I'm
>going to add a note to the feature tracker on sourceforge so that I
>don't forget to finish off this support for the 0.3 release.
>
>Cheers,
>
>Kal
>
>On Sat, 2004-08-28 at 18:25, Mathias Reinhardt wrote:
>  
>
>>Hello,
>>
>>I'm stuck with setting up TM4Web/Velocity 0.2rc1 with a hibernate
>>backend.
>>Configuration is
>>  <provider name="hibernate" 
>>factory="org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
>>    <provider-properties>
>>      <property name="hibernate.connection.url" 
>>value="jdbc:mysql://localhost/tm4mail"/>
>>      <property name="hibernate.connection.username" value="***"/>
>>      <property name="hibernate.connection.password" value="***"/>
>>      <property name="hibernate.dialect" 
>>value="net.sf.hibernate.dialect.MySQLDialect"/>
>>      <property name="hibernate.connection.driver_class" 
>>value="com.mysql.jdbc.Driver"/>
>>    </provider-properties>   
>>  </provider>
>>
>>I don't get any errors but I don't get any TopicMaps from by backend 
>>either. In stdout.log it says
>>"topicmaps element ignored for provider 'hibernate' which is not in-memory."
>>
>>Do I miss something or is Hibernate support currently not available with 
>>TM4Web/Velocity?
>>
>>Thanks,
>>Mathias
>>
>>
>>-------------------------------------------------------
>>This SF.Net email is sponsored by BEA Weblogic Workshop
>>FREE Java Enterprise J2EE developer tools!
>>Get your free copy of BEA WebLogic Workshop 8.1 today.
>>http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click
>>_______________________________________________
>>Tm4j-users mailing list
>>Tm4...@li...
>>https://lists.sourceforge.net/lists/listinfo/tm4j-users
>>    
>>

RE: [TM4J-users] exporting subject indicators

[Kal A. <ka...@te...>]

Hi Conal,

It definitely has to be a <subjectIndicatorRef> element. The <topicRef>
element won't add the subjectIndicator property to your topic. 

My advice would be to make it absolutely explicit in your input pipeline
that you are creating subject indicators and make full URIs:

<topic id="foo">
  <subjectIdentity>
    <subjectIndicatorRef
xlink:href="http://nztec.org/tm/harvest/xtm/nztec-ontology.xtm#foo"/>
  </subjectIdentity>
</topic>

It may seem like a bit of overkill, but its the most robust way to
ensure that what you mean is what is read by the XTM parser !

Cheers,

Kal


On Mon, 2004-09-27 at 08:52, Conal Tuohy wrote:
> Thanks for the prompt reply Kal! I'm really grateful for your help.
> 
> I'll bet that's it. No I don't have <subjectIndicatorRef> elements in my
> source.
> 
> I'm not using TMHarvest at present ... I'm using Cocoon pipelines with XSLT
> to harvest the topics. On the bright side, this makes it easy to add a
> subjectIndicator as a separate pipeline stage. If I understand you
> correctly, I would just need to add a <subjectIdentity> to my topics,
> pointing at themselves? It sounds a bit weird ... maybe I've misunderstood.
> 
> <topic id="foo"/>
> 
> would become
> 
> <topic id="foo">
> 	<subjectIdentity>
> 		<subjectIndicatorRef xlink:href="#foo"/>
> 	</subjectIdentity>
> </topic>
> 
> or maybe <topicRef> is better
> 
> <topic id="foo">
> 	<subjectIdentity>
> 		<topicRef xlink:href="#foo"/>
> 	</subjectIdentity>
> </topic>
> 
> Is that it?
> 
> Cheers
> 
> Con
> 
> > Do your input topic maps have <subjectIndicator> elements in
> > them ? (or
> > if you are harvesting them using TMHarvest, do you set the
> > subjectIndicator property) ? My guess is that you are only setting the
> > sourceLocator property. So your topic with the id pdf-file
> > gets a source
> > locator
> > http://www.nztec.org/tm/harvest/xtm/nztec-ontology.xtm#pdf-file,
> > but does not get a subjectIndicator. By default, the XTM
> > exporter tries
> > to be "clever" and if you export a TM with a base locator
> > http://foo.xtm
> > then all source locators of the format http://foo.xtm#bar get exported
> > as an ID attribute with the value bar.
> >
> > The merging works in memory because if topic A has
> > sourceLocator foo and
> > topic B has a subjectIndicator foo, then the topics must be
> > merged. When
> > you round trip to external files, however you will lose that
> > sourceLocator.
> >
> > My advice is to explicitly set the subject indicators in your ontology
> > topic map.
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
-- 
Kal Ahmed <ka...@te...>
techquila

RE: [TM4J-users] exporting subject indicators

[Conal T. <co...@pa...>]

Christoph Froehlich wrote:

> Am Mo, den 27.09.2004 schrieb Kal Ahmed um 9:26:
> > Hi Conal,
> >
> > Do your input topic maps have <subjectIndicator> elements
> in them ? (or
> > if you are harvesting them using TMHarvest, do you set the
> > subjectIndicator property) ? My guess is that you are only
> setting the
> > sourceLocator property.
>
> If its possible to do the merging and harvesting with tmnav,
> you should
> see the subject indicators either when using the section renderer or
> when using hypergraph/touchgraph renderer with the extended
> abstractor.
> The latter two display them with a green circle that is labeled "si"

Thanks Christoph. I've been using TMNav heavily :-) but I don't remember
seeing those subject indicators in TMNav. I think Kal is right that I need
to make the subject indicators explicit. I had a wrong idea based on the way
that TM4J sometimes adds subject indicators when exporting. I'll definitely
look out for it when I'm back at work tomorrow. This is good to know.

RE: [TM4J-users] exporting subject indicators

[Conal T. <co...@pa...>]

Thanks for the prompt reply Kal! I'm really grateful for your help.

I'll bet that's it. No I don't have <subjectIndicatorRef> elements in my
source.

I'm not using TMHarvest at present ... I'm using Cocoon pipelines with XSLT
to harvest the topics. On the bright side, this makes it easy to add a
subjectIndicator as a separate pipeline stage. If I understand you
correctly, I would just need to add a <subjectIdentity> to my topics,
pointing at themselves? It sounds a bit weird ... maybe I've misunderstood.

<topic id="foo"/>

would become

<topic id="foo">
	<subjectIdentity>
		<subjectIndicatorRef xlink:href="#foo"/>
	</subjectIdentity>
</topic>

or maybe <topicRef> is better

<topic id="foo">
	<subjectIdentity>
		<topicRef xlink:href="#foo"/>
	</subjectIdentity>
</topic>

Is that it?

Cheers

Con

> Do your input topic maps have <subjectIndicator> elements in
> them ? (or
> if you are harvesting them using TMHarvest, do you set the
> subjectIndicator property) ? My guess is that you are only setting the
> sourceLocator property. So your topic with the id pdf-file
> gets a source
> locator
> http://www.nztec.org/tm/harvest/xtm/nztec-ontology.xtm#pdf-file,
> but does not get a subjectIndicator. By default, the XTM
> exporter tries
> to be "clever" and if you export a TM with a base locator
> http://foo.xtm
> then all source locators of the format http://foo.xtm#bar get exported
> as an ID attribute with the value bar.
>
> The merging works in memory because if topic A has
> sourceLocator foo and
> topic B has a subjectIndicator foo, then the topics must be
> merged. When
> you round trip to external files, however you will lose that
> sourceLocator.
>
> My advice is to explicitly set the subject indicators in your ontology
> topic map.

Re: [TM4J-users] exporting subject indicators

[Christoph F. <cf...@fo...>]

Am Mo, den 27.09.2004 schrieb Kal Ahmed um 9:26:
> Hi Conal,
> 
> Do your input topic maps have <subjectIndicator> elements in them ? (or
> if you are harvesting them using TMHarvest, do you set the
> subjectIndicator property) ? My guess is that you are only setting the
> sourceLocator property. 

If its possible to do the merging and harvesting with tmnav, you should
see the subject indicators either when using the section renderer or
when using hypergraph/touchgraph renderer with the extended abstractor.
The latter two display them with a green circle that is labeled "si"


c



-- 
Christoph Froehlich <cf...@fo...>

Re: [TM4J-users] exporting subject indicators

[Kal A. <ka...@te...>]

Hi Conal,

Do your input topic maps have <subjectIndicator> elements in them ? (or
if you are harvesting them using TMHarvest, do you set the
subjectIndicator property) ? My guess is that you are only setting the
sourceLocator property. So your topic with the id pdf-file gets a source
locator http://www.nztec.org/tm/harvest/xtm/nztec-ontology.xtm#pdf-file,
but does not get a subjectIndicator. By default, the XTM exporter tries
to be "clever" and if you export a TM with a base locator http://foo.xtm
then all source locators of the format http://foo.xtm#bar get exported
as an ID attribute with the value bar.

The merging works in memory because if topic A has sourceLocator foo and
topic B has a subjectIndicator foo, then the topics must be merged. When
you round trip to external files, however you will lose that
sourceLocator.

My advice is to explicitly set the subject indicators in your ontology
topic map. 

Hope this helps!

Cheers,

Kal

On Mon, 2004-09-27 at 07:35, Conal Tuohy wrote:
> I recently switched from Hibernate to the In-Memory back-end and I now have a slight problem with persistence of subject indicators. With the In-Memory back-end you have to write the topic map to an XTM file to persist it. I THINK that all my subject indicators are in the topic map in memory, but are not written to my XTM, and this causes a problem next time I load the map. 
> 
> My TM is built by merging a bunch of XTM files which are harvested from various sources, and also including some ontological topicmaps. For instance I have an ontology file with a base-locator of http://www.nzetc.org/tm/harvest/xtm/nzetc-ontology.xtm which includes for instance a topic with id = "pdf-file". When I save my central (merged) XTM file I get the topic written out just the same way, as a topic with id="pdf-file", but with no<subjectIdentity>. Because the base locator of the merged file is different to the base locator of the ontology, I won't be able to refer to the same topic in a different file, using a full URI:
> http://www.nzetc.org/tm/harvest/xtm/nzetc-ontology.xtm#pdf-file
> ... because this subject indicator just isn't present in the (merged) XTM file I save.
> 
> I've looked at the source code of the TopicMapWalker and the XTMWriter and it looks like it should work, but I get nothing in my output.
> 
> So long as I have Cocoon running, and my topic map in memory, then I can merge topic maps into it and everything is fine; the topics merge as they should (this is why I think the subject indicators ARE present in memory), but as soon as I save the merged XTM and reload it, then my original subject indicators are gone and everything turns to custard.
> 
> Any ideas?
> 
> Con
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
-- 
Kal Ahmed <ka...@te...>
techquila

[TM4J-users] exporting subject indicators

[Conal T. <Con...@vu...>]

I recently switched from Hibernate to the In-Memory back-end and I now =
have a slight problem with persistence of subject indicators. With the =
In-Memory back-end you have to write the topic map to an XTM file to =
persist it. I THINK that all my subject indicators are in the topic map =
in memory, but are not written to my XTM, and this causes a problem next =
time I load the map.=20

My TM is built by merging a bunch of XTM files which are harvested from =
various sources, and also including some ontological topicmaps. For =
instance I have an ontology file with a base-locator of =
http://www.nzetc.org/tm/harvest/xtm/nzetc-ontology.xtm which includes =
for instance a topic with id =3D "pdf-file". When I save my central =
(merged) XTM file I get the topic written out just the same way, as a =
topic with id=3D"pdf-file", but with no<subjectIdentity>. Because the =
base locator of the merged file is different to the base locator of the =
ontology, I won't be able to refer to the same topic in a different =
file, using a full URI:
http://www.nzetc.org/tm/harvest/xtm/nzetc-ontology.xtm#pdf-file
... because this subject indicator just isn't present in the (merged) =
XTM file I save.

I've looked at the source code of the TopicMapWalker and the XTMWriter =
and it looks like it should work, but I get nothing in my output.

So long as I have Cocoon running, and my topic map in memory, then I can =
merge topic maps into it and everything is fine; the topics merge as =
they should (this is why I think the subject indicators ARE present in =
memory), but as soon as I save the merged XTM and reload it, then my =
original subject indicators are gone and everything turns to custard.

Any ideas?

Con

Re: [TM4J-users] TM4Web/Velocity 0.2rc1 supports Hibernate backends?

[Kal A. <ka...@te...>]

Hi Mathias,

I've checked in some changes for the TM4Web/Velocity package which
provides a basic level of support for listing the topic maps in a
database. You need a configuration like this:

<provider name="hibernate" 
factory="org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
    <provider-properties>
      <property name="hibernate.connection.url" 
value="jdbc:mysql://localhost/tm4mail"/>
      <property name="hibernate.connection.username" value="***"/>
      <property name="hibernate.connection.password" value="***"/>
      <property name="hibernate.dialect" 
value="net.sf.hibernate.dialect.MySQLDialect"/>
      <property name="hibernate.connection.driver_class" 
value="com.mysql.jdbc.Driver"/>
    </provider-properties>   
    <topicmaps template="topicmap.vm"/>
  </provider>

Note the extra <topicmaps/> element that is required. The template
attribute specifies the template to be used for the main topicmap page.
With this update, you are forced to always use topic.vm to render
topics, but I will work on adding support like you have in the normal
<topicmap> element to specify the template(s) to use to render topics.

In tmbrowse, all topic maps are listed on the index page with their
title as the full base URL of the topic map, so you end up with the URL
listed twice (because by default the index page shows the source URL as
well) - you may want to edit the listmaps.vm template to shorten this.

I hope this gets you the basic functionality you need right now - I'm
going to add a note to the feature tracker on sourceforge so that I
don't forget to finish off this support for the 0.3 release.

Cheers,

Kal

On Sat, 2004-08-28 at 18:25, Mathias Reinhardt wrote:
> Hello,
> 
> I'm stuck with setting up TM4Web/Velocity 0.2rc1 with a hibernate
> backend.
> Configuration is
>   <provider name="hibernate" 
> factory="org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
>     <provider-properties>
>       <property name="hibernate.connection.url" 
> value="jdbc:mysql://localhost/tm4mail"/>
>       <property name="hibernate.connection.username" value="***"/>
>       <property name="hibernate.connection.password" value="***"/>
>       <property name="hibernate.dialect" 
> value="net.sf.hibernate.dialect.MySQLDialect"/>
>       <property name="hibernate.connection.driver_class" 
> value="com.mysql.jdbc.Driver"/>
>     </provider-properties>   
>   </provider>
> 
> I don't get any errors but I don't get any TopicMaps from by backend 
> either. In stdout.log it says
> "topicmaps element ignored for provider 'hibernate' which is not in-memory."
> 
> Do I miss something or is Hibernate support currently not available with 
> TM4Web/Velocity?
> 
> Thanks,
> Mathias
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by BEA Weblogic Workshop
> FREE Java Enterprise J2EE developer tools!
> Get your free copy of BEA WebLogic Workshop 8.1 today.
> http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
-- 
Kal Ahmed <ka...@te...>
techquila

RE: [TM4J-users] differently-scoped names causing a merge?

[Kal A. <ka...@te...>]

Hi Conal,

Thanks for the update. I agree that XTMBuilder should either flag up a
misplaced <scope> element as an error or process it, but it should
definitely not just silently ignore it. There is an optional flag which
turns on XTM validation: http://www.tm4j.org/tm4j/xtmbuilder/validation
which you set by calling the setProperty method of XTMBuilder (although
this gets hidden from you if you use the SerializedTopicMapSource
helper).

Following the maxim "Be strict in what you produce and generous in what
you accept", I think that the builder should be able to handle a <scope>
element appearing in the wrong place as a child of <baseName>,
<occurrence> and <association> - so I'll enter a bug for this into the
tracker.

Cheers,

Kal

On Sun, 2004-09-26 at 08:06, Conal Tuohy wrote:
> I think I've found what was going wrong with my topicmaps (in which the name
> scopes were being ignored).
> 
> I debugged org.tm4j.topicmap.utils.XTMBuilder and found that for a baseName
> to be scoped, the scope element has to appear before the baseNameString
> element. Looking at the DTD, this is of course perfectly correct - the
> optional <scope> child of a <baseName> must be the first child element. So
> it was my mistake. D'oh! :-(
> 
> But it seems to me that it'd be nice for XTMBuilder to flag a <scope>
> appearing after a <baseNameString> as an error, rather than ignoring it. The
> scoping of <occurrence> and <association> elements is handled differently,
> and for these elements XTMBuilder seems to accept <scope> elements which
> appear (invalidly) after a <resourceData>, <member>, etc,
> 
> Cheers
> 
> Con
> 
> 
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
-- 
Kal Ahmed <ka...@te...>
techquila

RE: [TM4J-users] differently-scoped names causing a merge?

[Conal T. <co...@pa...>]

I think I've found what was going wrong with my topicmaps (in which the name
scopes were being ignored).

I debugged org.tm4j.topicmap.utils.XTMBuilder and found that for a baseName
to be scoped, the scope element has to appear before the baseNameString
element. Looking at the DTD, this is of course perfectly correct - the
optional <scope> child of a <baseName> must be the first child element. So
it was my mistake. D'oh! :-(

But it seems to me that it'd be nice for XTMBuilder to flag a <scope>
appearing after a <baseNameString> as an error, rather than ignoring it. The
scoping of <occurrence> and <association> elements is handled differently,
and for these elements XTMBuilder seems to accept <scope> elements which
appear (invalidly) after a <resourceData>, <member>, etc,

Cheers

Con

Re: [TM4J-users] TM4Web/Velocity 0.2rc1 supports Hibernate backends?

[Kal A. <ka...@te...>]

Hi Mathias,

I'm afraid that I don't have a release plan at the moment - I'm trying 
to get TMAPI 1.0 and TM4J 1.0 finished over the next week or so. If I 
can get a bit of time (and I might have some time this weekend), I'll 
take a look at implementing this - I just can't promise anything right 
now. I'll keep your need in mind though!

Cheers,

Kal

Mathias Reinhardt wrote:

> Hi Kal,
>
> I need the function to fetch the available topic maps from
> a Hibernate backend quite urgent for a project. Do you have
> a release plan for 0.3?
> |
> Greetings,
> Mathias
>
>
> |Kal Ahmed wrote:
>
>> Hi Mathias,
>>
>> TM4Web/Velocity does support the Hibernate backend, but the <topicmaps>
>> element is only provided for loading topic maps from a file-system
>> directory (so its only used with the in-memory backend). You have to
>> configure each topic map you want displayed separately. It would
>> probably be nice if there was an option to list either all of the topic
>> maps in the database or some subset (e.g. by matching the baseLocator
>> against a regular expression). I'll look into adding something like that
>> for 0.3.
>>
>> Cheers,
>>
>> Kal
>>
>> On Sat, 2004-08-28 at 18:25, Mathias Reinhardt wrote:
>>  
>>
>>> Hello,
>>>
>>> I'm stuck with setting up TM4Web/Velocity 0.2rc1 with a hibernate
>>> backend.
>>> Configuration is
>>>  <provider name="hibernate" 
>>> factory="org.tm4j.topicmap.hibernate.TopicMapProviderFactoryImpl">
>>>    <provider-properties>
>>>      <property name="hibernate.connection.url" 
>>> value="jdbc:mysql://localhost/tm4mail"/>
>>>      <property name="hibernate.connection.username" value="***"/>
>>>      <property name="hibernate.connection.password" value="***"/>
>>>      <property name="hibernate.dialect" 
>>> value="net.sf.hibernate.dialect.MySQLDialect"/>
>>>      <property name="hibernate.connection.driver_class" 
>>> value="com.mysql.jdbc.Driver"/>
>>>    </provider-properties>    </provider>
>>>
>>> I don't get any errors but I don't get any TopicMaps from by backend 
>>> either. In stdout.log it says
>>> "topicmaps element ignored for provider 'hibernate' which is not 
>>> in-memory."
>>>
>>> Do I miss something or is Hibernate support currently not available 
>>> with TM4Web/Velocity?
>>>
>>> Thanks,
>>> Mathias
>>>
>>>
>>> -------------------------------------------------------
>>> This SF.Net email is sponsored by BEA Weblogic Workshop
>>> FREE Java Enterprise J2EE developer tools!
>>> Get your free copy of BEA WebLogic Workshop 8.1 today.
>>> http://ads.osdn.com/?ad_id=5047&alloc_id=10808&op=click
>>> _______________________________________________
>>> Tm4j-users mailing list
>>> Tm4...@li...
>>> https://lists.sourceforge.net/lists/listinfo/tm4j-users
>>>   
>>

[TM4J-users] differently-scoped names causing a merge?

[Conal T. <Con...@vu...>]

Is this a bug in tm4j? Or what am I doing wrong?

In this example I have 2 topic maps, each containing a single topic, =
with the same baseName ("name"), but in different scopes. All the topic =
characteristics in each topicmap are scoped with a topic reifying the =
topicmap itself, so the name-spaces should be distinct, and I should get =
no name-based merging. But when I load the 2 maps into a 3rd (empty) =
map, I find that the 2 topics merge. When I serialise the resulting map, =
I see a single topic with 2 identical, and unscoped(!) baseNames.=20

The 2 input topic maps:

<?xml version=3D"1.0" encoding=3D"UTF-8"?>
<topicMap xmlns=3D"http://www.topicmaps.org/xtm/1.0/" =
xmlns:xlink=3D"http://www.w3.org/1999/xlink">
	<topic id=3D"topic-map-source">
		<subjectIdentity>
			<resourceRef =
xlink:href=3D"http://www.nzetc.org/tm/harvest/xtm/merge-test-1.xtm"/>
		</subjectIdentity>
	</topic>

	<topic id=3D"t1">
		<baseName>
			<baseNameString>name</baseNameString>
			<scope>
				<topicRef xlink:href=3D"#topic-map-source"/>
			</scope>
		</baseName>
		<occurrence>
			<resourceData>topic 1</resourceData>
			<scope>
				<topicRef xlink:href=3D"#topic-map-source"/>
			</scope>
		</occurrence>
	</topic>

 </topicMap>

=20
<?xml version=3D"1.0" encoding=3D"UTF-8"?>
<topicMap xmlns=3D"http://www.topicmaps.org/xtm/1.0/" =
xmlns:xlink=3D"http://www.w3.org/1999/xlink">

	<topic id=3D"topic-map-source">
		<subjectIdentity>
			<resourceRef =
xlink:href=3D"http://www.nzetc.org/tm/harvest/xtm/merge-test-2.xtm"/>
		</subjectIdentity>
	</topic>

	<topic id=3D"t2">
		<baseName>
			<baseNameString>name</baseNameString>
			<scope>
				<topicRef xlink:href=3D"#topic-map-source"/>
			</scope>
		</baseName>
		<occurrence>
			<resourceData>topic 2</resourceData>
			<scope>
				<topicRef xlink:href=3D"#topic-map-source"/>
			</scope>
		</occurrence>
	</topic>

 </topicMap>

The merged topic map:

<?xml version=3D"1.0" encoding=3D"UTF-8"?>
<topicMap xmlns=3D"http://www.topicmaps.org/xtm/1.0/" =
xmlns:xlink=3D"http://www.w3.org/1999/xlink" id=3D"x1k3d9d52r-0" =
xml:base=3D"file:/C:/Program Files/Apache Tomcat =
4.0/webapps/ROOT/nzetc/tm/export/nzetc.xtm">
  <topic id=3D"x1k3d9d5eq-2">
    <subjectIdentity>
      <resourceRef =
xlink:href=3D"http://www.nzetc.org/tm/harvest/xtm/merge-test-2.xtm"/>
    </subjectIdentity>
  </topic>
  <topic id=3D"topic-map-source">
    <subjectIdentity>
      <resourceRef =
xlink:href=3D"http://www.nzetc.org/tm/harvest/xtm/merge-test-1.xtm"/>

    </subjectIdentity>
  </topic>
  <topic id=3D"t1">
    <baseName id=3D"x1k3d9d5eq-0">
      <baseNameString>name</baseNameString>
    </baseName>
    <baseName id=3D"x1k3d9d5eq-3">
      <baseNameString>name</baseNameString>

    </baseName>
    <occurrence id=3D"x1k3d9d5eq-1">
      <scope>
        <topicRef xlink:href=3D"#topic-map-source"/>
      </scope>
      <resourceData>topic 1</resourceData>
    </occurrence>
    <occurrence id=3D"x1k3d9d5eq-4">

      <scope>
        <topicRef xlink:href=3D"#x1k3d9d5eq-2"/>
      </scope>
      <resourceData>topic 2</resourceData>
    </occurrence>
  </topic>
</topicMap>

I'm running tm4j 0.9.5 with the In-Memory Provider.=20

The 2 topicmaps are merged into my central topicmap using =
provider.addTopicMap(xtmSource, existingMap)

I haven't just disabled the TNC because I sometimes DO want to merge =
using names.

I'd appreciate any comments! Thanks

Con

Re: [TM4J-users] SOLVED: Basename Merging

[Christoph F. <cf...@fo...>]

Hi

for the records: I just found out how to influence merging properties
with TMNav.

As Stefan stated, setting the property on the command line does not
work.

Instead you need to set the properties on a per-provider-base in
tmnav.xml (located in the tmnav-settings directory). The following is an
example that sets all the properties to their default values:

    <provider id=3D"inmemory-provider">
    =20
<factoryclassname>org.tm4j.topicmap.memory.TopicMapProviderFactoryImpl</fac=
toryclassname>
      <displayname>In-Memory-Provider</displayname>
      <description>A provider that stores topicmaps in
memory</description>
      <!-- influence merging -->=09
      <property key=3D"tm4j.make.consistent">false</property>
      <property key=3D"tm4j.merge.referenced.maps">false</property>
      <property key=3D"tm4j.name.based.merge">false</property>
      <property key=3D"tm4j.static.merge">false</property>
      =09
    </provider>


Hope that helps
c


Am Mi, den 14.07.2004 schrieb Stefan Lischke um 18:11:
> Hi Andrea,
>=20
> I'm also using TM4J without baseName merging. You can turn it off by=20
> simply changing the System.property :
>=20
> tm4j.name.based.merge=3Dfalse
>=20
> If you use the merging skript, just add it to the java call with
> -Dtm4j.name.based.merge=3Dfalse
>=20
> But as i stated in an earlier mail, you cannot use this when working
> with TMNav, so you can't visualize TopicMaps which contain Topics with
> same BaseNames.
>=20
> Stefan
>=20
>=20
> Andrea Carradori wrote:
>=20
> > Hi,
> >
> > I am looking for a way to deactivate the automatic merging of topics
> > basing on basenames. I want two topics to merge only if they have
> the=20
> > same Subject Indicator: it could be that, even if two topics have
> the=20
> > same basename, they mean different things=85
> >
> > I=92m using a Hibernate back end.
> >
> > Thanks for your help!
> >
> > Andrea
> >
>=20
>=20
>=20
> -------------------------------------------------------
> This SF.Net email sponsored by Black Hat Briefings & Training.
> Attend Black Hat Briefings & Training, Las Vegas July 24-29 -=20
> digital self defense, top technical experts, no vendor pitches,=20
> unmatched networking opportunities. Visit www.blackhat.com
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
--=20
Christoph Froehlich <cf...@fo...>

Re: [TM4J-users] Hibernate Backend and thread safety

[Kal A. <ka...@te...>]

Hi Bent,

Are your users accessing the topic map for writes as well as reads ?
Currently the TM4J backends are not coded for thread-safety and so are
almost certainly not thread-safe. If your application can instead
serialize access to the topic map, that is the safest way to go at the
moment. Support for multiple threads is definitely a feature I want to
support, but my feeling is that support will have to wait until 1.0 has
been released.

Cheers,

Kal

On Thu, 2004-09-16 at 13:16, Bent Andre Solheim wrote:
> Hi,
> 
> I'm using the Hibernate backend in a web application and keep getting  
> error messages when several users are accessing the application  
> simultaneously. I'm using one shared TopicMap object for the entire  
> application. I'm worried about thread safetly of the TopicMap object.  
> Should I use one TopicMap object per request/thread?
> 
> Best regards
> Bent
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM.
> Deadline: Sept. 24. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Tm4j-users mailing list
> Tm4...@li...
> https://lists.sourceforge.net/lists/listinfo/tm4j-users
-- 
Kal Ahmed <ka...@te...>
techquila

RE: [TM4J-users] Hibernate Backend and thread safety

[Andy P. <an...@il...>]

> I'm using the Hibernate backend in a web application and keep getting  
> error messages when several users are accessing the application  
> simultaneously. I'm using one shared TopicMap object for the entire  
> application. I'm worried about thread safetly of the TopicMap object.  
> Should I use one TopicMap object per request/thread?

Yes. 

Take a look at http://www.hibernate.org/43.html if you haven't already.


Andy.

[TM4J-users] Hibernate Backend and thread safety

[Bent A. S. <be...@us...>]

Hi,

I'm using the Hibernate backend in a web application and keep getting  
error messages when several users are accessing the application  
simultaneously. I'm using one shared TopicMap object for the entire  
application. I'm worried about thread safetly of the TopicMap object.  
Should I use one TopicMap object per request/thread?

Best regards
Bent

Re: [TM4J-users] Re: unable to open topic maps in LTM format

[Kal A. <ka...@te...>]

The encoding directive and the #TOPICMAP directive should be implemented
(if they aren't its a bug). The #INCLUDE directive is not part of the
version of LTM that TM4J currently supports. Ontopia added this feature
recently - this is one of the problems of supporting a non-standard file
format...there is not much notification of these kind of changes.

Cheers,

Kal

On Wed, 2004-09-15 at 08:26, Christoph Froehlich wrote:
> Hi Vidar
> 
> Am Di, den 14.09.2004 schrieb Vidar Gundersen um 19:55:
> > ===== Opprinnelig melding fra Christoph Froehlich | Tue, 14 Sep 2004:
> > > If you are sure, would you mind to send me one the maps that tmnav
> > > wasn't able to read?
> > 
> > ===== Original message from SourceForge net | Tue, 14 Sep 2004:
> > > It seems to me that tm4j currently does not support the 
> > > # (line) comments:
> > > The first three lines of the offending map:
> > > @"utf-8"
> > > #TOPICMAP jills-tm
> > > #INCLUDE "dc.ltmm"
> > 
> > i guess # are not line comments, but processing directives:
> > http://www.ontopia.net/download/ltm.html#N227
> > 
> Ah, sorry, I haven't read the specs (carefully).
> 
> Nevertheless it seems to me, that tm4j does not implement them.
> I think we should mention this in the docs.
> 
> 
> 
> bye
> c
> 
> > 
> > kind regards,
> > Vidar Gundersen
-- 
Kal Ahmed <ka...@te...>
techquila