On 02/10/18 07:37, Nouwt, B. (Barry) wrote:
Hi Andy, thanks for the extended reply.
It's unclear to me, from your description below, whether the "full" server
(that starts via the bash script) also uses the web.xml when starting? Since this web.xml
file is normally only used for deployment on a servlet container (like tomcat).
Yes, it does.
The standalone packaging of the full server is jetty+the webapp and does
normal webapp startup processing web.xml.
Andy
Anyway, if I understand you correctly, my only option is adding custom code to
the end of FusekiServerListener.serverInitialization(). Although not ideal, for
demo purposes this suffices.
Thanks for your help!
Barry
B. (Barry) Nouwt
Medior Innovator Semantic Technology
Connected Business
T +31 (0)88 866 56 91
M +31 (0)64 977 53 56
E [email protected]
This message may contain information that is not intended for you. If you are
not the addressee or if this message was sent to you by mistake, you are
requested to inform the sender and delete the message. TNO accepts no liability
for the content of this e-mail, for the manner in which you use it and for
damage of any kind resulting from the risks inherent to the electronic
transmission of messages.
-----Original Message-----
From: Andy Seaborne <[email protected]>
Sent: zondag 30 september 2018 11:56
To: [email protected]
Subject: Re: get derivations via Fuseki
Hi Barry,
If you can use the embedded server, it is easier to setup with extended
functionality, as per JENA-1435, but then no UI and admin
In Jena 3.9.0 the embedded server is renamed as "main", and the UI/admin/core engine server is the
"full" or "webapp" server [*]
The mechanisms of JENA-1435 will work in the full server (even while
running!) but I can't think of a way to get the ServletContext, which is used
to store the Fuseki configuration, which isn't either a code change or a change
to web.xml.
The code change is to FusekiServerListener.serverInitialization. At the end of
that method, call your code. (For a web.xml chnage, add your own
ServletContextListener to after Fuseki's).
That gets the ServletContext; from there get the ServiceDispatchRegistry
ServiceDispatchRegistry registry =
ServiceDispatchRegistry.get(ServletContext servletContext)
and then manipulate registry to add your service.
See around FusekiServer.java#L474 for manipulating the ServiceDispatchRegistry.
https://github.com/apache/jena/blob/aef35ee50eef9e0820d23d5ba184906b177a5c15/jena-fuseki2/jena-fuseki-main/src/main/java/org/apache/jena/fuseki/main/FusekiServer.java#L474
For the full server, you need to put your code onto the classpath of the server so
instead of running the standalone server jar with "-jar"
(which is what the script does), you need to use the combined fuseki-server.jar
jar ad one element of the classpath and your
code+extra dependencies then call "main":
java -cp fuseki-server-mod.jar:YourCode.jar:other.jar \
org.apache.jena.fuseki.cmd.FusekiCmd "$@"
If you use the WAR file, you'll need to unpack and repack it with your code in
it.
If there is a way to get the ServletContext, "ja:loadClass" would work.
However, the change in JNA-1435 is to go from a global registry to a one
per ServletContext (allows for multiple servers in one JVM).
Or run a modified Fuseki with call to your code in
FusekiServerListener.serverInitialization.
---
http://jena.apache.org/documentation/fuseki2/fuseki-embedded.html
src/test/java/org/apache/jena/fuseki/embedded/examples/ExtendFuseki_AddService_1.java
---
Andy
[*] Because of the way webapps start up in servlet containers like
Tomcat, the two servers have different initialization code.
The difference in the two forms comes down to the fact that the full
server has state on disk (the run/ area) where it manages datasets, and
the main server is given datasets to publish.
Some of the UI applies to the "main" server and could be shared, some
does not (e.g. adding datasets). Something for the future maybe.
I'm currently engaged in a project deploying Fuseki into a cloud
environment as a (micro)service, deployed by automation etc etc.
No UI; it's accessed by other services.
On 28/09/18 15:41, Nouwt, B. (Barry) wrote:
Hi all,
I would like to include the derivation service I describe below in the
standalone version of Fuseki (i.e. started by calling the ./fuseki-server
script). Is this possible? I assume not, since in this JIRA-1435 issue
(https://issues.apache.org/jira/browse/JENA-1435), I read:
"At least initially, this extensibility is only supported for the programmatic
Fuseki server and have a programmatic configuration API."
Can anyone confirm it is not possible to load my custom service if I use the
fuseki-server script? Or can I maybe use "ja:loadClass" from a
configuration.ttl file to load it?
Thanks for the help!
Regards, Barry
-----Original Message-----
From: Nouwt, B. (Barry) <[email protected]>
Sent: maandag 28 mei 2018 09:49
To: [email protected]
Subject: RE: get derivations via Fuseki
Hi Andy,
Thanks for the clarification.
Regarding "What is also good to know is whether Jena has the right extension
points." I expect some modifications to the Apache Jena source are required. See my
previous mail where I have some remarks/questions:
- The Derivation interface does not allow inspecting the derivation via code
for a particular triple. It only declares a toString() and printTrace() method.
To be able to generate the derivation log in a JSON or RDF format, this
interface would probably need to be adapted (or we would need to cast a
Derivation object to some implementation of this interface making the code less
reasoner independent).
- The InfGraph interface does not allow checking whether derivation logging is
enabled. You can only enable/disable it using the above setDerivationLogging()
method. There is a method in BasicForwardRuleInfGraph.shouldLogDerivations(),
though. Is there some other way to check whether it is enabled? Or should
InfGraph get an additional method?
- Since the derivations can take up quite some memory, you need to enable
derivation logging on an InfGraph using the setDerivationLogging(true) method.
Can the DerivationService that I'm building call this method on the InfGraph
when the Fuseki dataset is loaded? Or would it be better to modify the InfGraph
Assembler code to configure it from a Fuseki configuration file? In that case a
user would need to remember to enable derivation logging, when it wants to use
the derivation service.
Regards,
Barry
B. (Barry) Nouwt
Medior Innovator Semantic Technology
Connected Business
T +31 (0)88 866 56 91
M +31 (0)64 977 53 56
E [email protected]
This message may contain information that is not intended for you. If you are
not the addressee or if this message was sent to you by mistake, you are
requested to inform the sender and delete the message. TNO accepts no liability
for the content of this e-mail, for the manner in which you use it and for
damage of any kind resulting from the risks inherent to the electronic
transmission of messages
-----Original Message-----
From: Andy Seaborne <[email protected]>
Sent: zondag 27 mei 2018 17:29
To: [email protected]
Subject: Re: get derivations via Fuseki
On 17/05/18 10:11, Nouwt, B. (Barry) wrote:
Hi all,
I've successfully modified the TestFusekiCustomOperation example to a working
DerivationService test. I can request the derivation of some triple using the
following request:
http://localhost:3030/ds/derivation?triple=(<http://tno.nl/B>
<http://tno.nl/departure> "later")
and it returns the derivation log text:
Rule calcETD concluded (<http://tno.nl/B> <http://tno.nl/departure> 'later') <-
Fact (<http://tno.nl/A> rdf:type <http://tno.nl/Ship>)
Rule calcETA concluded (<http://tno.nl/A> <http://tno.nl/arrival> 'late')
<-
Fact (<http://tno.nl/A> rdf:type <http://tno.nl/Ship>)
Fact (<http://tno.nl/A> <http://tno.nl/speed> 'Slow')
Fact (<http://tno.nl/B> rdf:type <http://tno.nl/Truck>)
Fact (<http://tno.nl/B> <http://tno.nl/picksUpFrom>
<http://tno.nl/A>)
Of course, the questions mentioned in an earlier mail remain, but at least I
know it's possible. Next, I will make it more user friendly and generic so we
can include it in our semantic platform.
@Andy: is the Apache Jena project interested in getting the code when I have a
more generic version? Or would it be better to keep the code separate from
Apache Jena? If Apache Jena is interested, I should probably investigate how
this process works (since I have no experience yet with contributing to Apache
Jena projects).
Barry - the project is always interested and it doesn't sound too big.
What is also good to know is whether Jena has the right extension points.
General comments on additional functionality:
What any Apache project isn't, by default, is a place to transfer code for
future maintenance by the project team. If some on the team are interested
great, but it is better to know there is interest. Any significant sized
contribution will need some degree of on-going interest to keep it maintained.
By labelling a Github project (e.g.) "Powered by Apache Jena", search engines
will be able to find it. Not everything has to be part of one humongous build.
Andy
Regards, Barry
-----Original Message-----
From: Nouwt, B. (Barry) <[email protected]>
Sent: dinsdag 15 mei 2018 17:55
To: [email protected]
Subject: RE: get derivations via Fuseki
Hi Andy, thanks for your reply.
I took a look at the TestFusekiCustomOperation example and it looks quite
straightforward. I started with a first implementation and the following
questions arised:
- The Derivation interface does not allow inspecting the derivation via code
for a particular triple. It only declares a toString() and printTrace() method.
To be able to generate the derivation log in a JSON or RDF format, this
interface would probably need to be adapted (or we would need to cast a
Derivation object to some implementation of this interface making the code less
reasoner independent).
- The InfGraph interface does not allow checking whether derivation logging is
enabled. You can only enable/disable it using the above setDerivationLogging()
method. There is a method in BasicForwardRuleInfGraph.shouldLogDerivations(),
though. Is there some other way to check whether it is enabled? Or should
InfGraph get an additional method?
- Since the derivations can take up quite some memory, you need to enable
derivation logging on an InfGraph using the setDerivationLogging(true) method.
Can the DerivationService that I'm building call this method on the InfGraph
when the Fuseki dataset is loaded? Or would it be better to modify the InfGraph
Assembler code to configure it from a Fuseki configuration file? In that case a
user would need to remember to enable derivation logging, when it wants to use
the derivation service.
So, I think some modifications to the Apache Jena source are required to enable
the derivation service feature.
Regards,
Barry
-----Original Message-----
From: Andy Seaborne <[email protected]>
Sent: donderdag 10 mei 2018 14:48
To: [email protected]
Subject: Re: get derivations via Fuseki
Hi Barry,
On 08/05/18 16:53, Nouwt, B. (Barry) wrote:
Hello everyone,
I know Apache Jena allows a developer to access the derivation of an inferred
triple using the InfModel.getDerivation() method. Can I also access this
explanation via Apache Jena Fuseki GUI?
I have configured Apache Jena Fuseki in such a way that the GenericRuleReasoner
infers new triples based on my dataset and a few rules. I am able to fire a
SPARQL query via Fuseki GUI and the answer includes the inferred triples as
expected. For our project, we would like to access the derivation of an
inferred triple via the GUI (as is possible in, for example, Protégé).
* Is this already possible in Fuseki? If not;
* Would it be interesting if we contribute to Fuseki to add such a
feature? And if so;
* Are there any relevant pointers to get us started?
For example, I noticed (in the 3.7.0 release) the following JIRA that sounds
like a starting point:
JENA-1435: Provide extensibility of Fuseki with new services.
It is now possible to add custom services to a Fuseki service, not
just the services provided by the Fuseki distribution.
Source:
https://jena.markmail.org/search/?q=3.7.0#query:3.7.0+page:1+mid:opdc
b
i6qhrim4bsv+state:results
If it is indeed possible to extend Fuseki with custom services, it might be
possible to introduce a new service called 'derivation' that, given a triple,
produces the derivation log for that triple. It would produce the explanation
in some JSON format. The Apache Jena Fuseki GUI could parse this result and
show it on screen.
So, my first question is: is it already possible to retrieve the derivation log
using Apache Jena Fuseki and if so, how?
There isn't such a feature.
My follow-up question: would it be interesting if we could contribute such a
feature to Fuseki?
Yes - and it should be possible to develop such a feature without needing to
modifying the Fuseki source. That's the point of JENA-1435.
There are some tests in TestFusekiCustomOperation
It is easier to develop functionality using the programmatic, same-JVM form of
the Fuseki server because the FusekiServer.Builder.
Andy
Regards,
Barry
This message may contain information that is not intended for you. If you are
not the addressee or if this message was sent to you by mistake, you are
requested to inform the sender and delete the message. TNO accepts no liability
for the content of this e-mail, for the manner in which you use it and for
damage of any kind resulting from the risks inherent to the electronic
transmission of messages.
