ng4j-namedgraphs — Discussion of Named Graphs in RDF

Re: [namedgraphs] Updating Libraries

[Olaf H. <ha...@in...>]

Hey,

On Saturday 25 July 2009 17:49:56 Tim Pizey wrote:
> Hi Olaf,
>
> On Saturday 25 July 2009 14:04:51 Olaf Hartig wrote:
> > Hello,
> >
> > I'm a bit unsure. I want to upgrade the included nekohtml.jar to the
> > latest version. Can I simply replace the old version by the latest one
> > (maybe with a name that reflects the version number) and update
> > ./lib/LIBRARIES.txt ? Or will this break something for Maven users?
>
> The Maven build does not use any of the files in the /lib/ directory, so
> this will have no effect.

Okay.

> Interestingly this is not a jar mentioned in the pom.xml so I assume it is
> a transitive dependency.

 ./lib/LIBRARIES.txt says it is included with GRDDL 0.3. However, it is likely 
that we will remove grddl.jar before the next release. nekohtml.jar will 
still be used.

Olaf

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Tim P. <tim...@ou...>]

Hi Olaf, 

On Saturday 25 July 2009 12:40:59 Olaf Hartig wrote:
> Hey Tim,
> 
> On Friday 24 July 2009 12:10:41 Tim Pizey wrote:
> > [...]
> > I have added the patch to the tracker, but also attach here.
> 
> The patch works with Jena 2.5.7. The change in
>    tests/de/fuberlin/wiwiss/ng4j/sparql/MyResultSet.java
> is not necessary.

I will check when I am back at my office, but all I was doing was what Eclipse told me, 
that the method signature had now changed to QuerySolution.
 
> When you edit a source file and want to commit it please respect and keep the 
> indentation style in this file, also for new methods. With 
> NamedGraphDataset.close you didn't do so.

OK, please create and commit an Eclipse style file, if you use Eclipse, 
or I will and will check that I have the style rules correct. 

Maybe I should add a Checkstyle file?

> > What I have not done is changed any jars in /lib.
> > Those changes would be to update antlr to version 2.7.7
> > and replace jena.jar and jenatest.jar with jena 2.5.7 jars
> 
> Don't forget arq.jar

Good catch.

> > The bouncycastle jars should not be different, we are just using them
> > directly fromn the maven repository now.
> 
> What do you mean by "we" here?
> Remember, not everyone is using Maven (e.g. me). Maven shouldn't be a 
> requirement to compile NG4J.

I have patched the Maven pom.xml to use the Maven central repository files, 
this will have no effect on the ant build, the actual jar files, in our lib and 
the maven central repository, I believe to be the same. 

> > I don't know if you are happy to continue to deploy to Melati.org or
> > whether there is a desire to get the ng4j jars into the main maven repos.
> > It is actually a bit of work to get them in and keep them updated.
> > (See the bouncy castle process)
> 
> ??? I was not aware that I/we "deploy to Melati.org." I do not even know what 
> Melati.org is.
> 
> Seriously, I cannot answer this question. You are the Maven expert. What do 
> you say?

OK, what I am asking is do we, like BouncyCastle and Jena, wish to go to 
the effort of trying to get our jars into the Maven central repository, this 
is a bit of work. I do publish our jar to am open repository which I maintain 
for my own use: http://melati.org/maven2, this is hard coded into the pom.xml file.
Unless others are interested I think we should leave this for now.

> > My sourceforge id is timp
> 
> I added you as a developer to the NG4J SourceForge project. This means you 
> have write access to CVS now. Welcome!

Thank you, I will be careful. 

yours
Tim 

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Olaf H. <ha...@in...>]

Hey,

On Saturday 25 July 2009 17:39:45 Tim Pizey wrote:
> Hi Olaf,
>
> On Saturday 25 July 2009 12:40:59 Olaf Hartig wrote:
> > [...]
> > When you edit a source file and want to commit it please respect and keep
> > the indentation style in this file, also for new methods. With
> > NamedGraphDataset.close you didn't do so.
>
> OK, please create and commit an Eclipse style file, if you use Eclipse,
> or I will and will check that I have the style rules correct.

Sorry, I don't use Eclipse.

> [...]
> OK, what I am asking is do we, like BouncyCastle and Jena, wish to go to
> the effort of trying to get our jars into the Maven central repository,
> this is a bit of work. I do publish our jar to am open repository which I
> maintain for my own use: http://melati.org/maven2, this is hard coded into
> the pom.xml file. Unless others are interested I think we should leave this
> for now.

Okay, I'm fine with that.

Greetings,
Olaf

Re: [namedgraphs] Updating Libraries

[Tim P. <tim...@ou...>]

Hi Olaf, 

On Saturday 25 July 2009 14:04:51 Olaf Hartig wrote:
> Hello,
> 
> I'm a bit unsure. I want to upgrade the included nekohtml.jar to the latest 
> version. Can I simply replace the old version by the latest one (maybe with a 
> name that reflects the version number) and update ./lib/LIBRARIES.txt ? Or 
> will this break something for Maven users?

The Maven build does not use any of the files in the /lib/ directory, so this will have no effect. 

Interestingly this is not a jar mentioned in the pom.xml so I assume it is a transitive dependency. 

cheers
Tim

[namedgraphs] Updating Libraries

[Olaf H. <ha...@in...>]

Hello,

I'm a bit unsure. I want to upgrade the included nekohtml.jar to the latest 
version. Can I simply replace the old version by the latest one (maybe with a 
name that reflects the version number) and update ./lib/LIBRARIES.txt ? Or 
will this break something for Maven users?

Greetings,
Olaf

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Olaf H. <ha...@in...>]

Hey Tim,

On Friday 24 July 2009 12:10:41 Tim Pizey wrote:
> [...]
> I have added the patch to the tracker, but also attach here.

The patch works with Jena 2.5.7. The change in
   tests/de/fuberlin/wiwiss/ng4j/sparql/MyResultSet.java
is not necessary.

When you edit a source file and want to commit it please respect and keep the 
indentation style in this file, also for new methods. With 
NamedGraphDataset.close you didn't do so.

> What I have not done is changed any jars in /lib.
> Those changes would be to update antlr to version 2.7.7
> and replace jena.jar and jenatest.jar with jena 2.5.7 jars

Don't forget arq.jar

> The bouncycastle jars should not be different, we are just using them
> directly fromn the maven repository now.

What do you mean by "we" here?
Remember, not everyone is using Maven (e.g. me). Maven shouldn't be a 
requirement to compile NG4J.

> I don't know if you are happy to continue to deploy to Melati.org or
> whether there is a desire to get the ng4j jars into the main maven repos.
> It is actually a bit of work to get them in and keep them updated.
> (See the bouncy castle process)

??? I was not aware that I/we "deploy to Melati.org." I do not even know what 
Melati.org is.

Seriously, I cannot answer this question. You are the Maven expert. What do 
you say?

> My sourceforge id is timp

I added you as a developer to the NG4J SourceForge project. This means you 
have write access to CVS now. Welcome!

Greetings,
Olaf

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Tim P. <tim...@ou...>]

Hi Olaf, 

On Thursday 23 Jul 2009 11:41:33 Olaf Hartig wrote:
> On Wednesday 22 July 2009 14:35:55 Tim Pizey wrote:
> > [...]
> > I now have everything building with 2.5.7 and with the BouncyCastle jars
> > from central repository.
> 
> That's cool.
> 
> > > I suspect that there will be a bit of work here, rather than submitting
> > > patches I would prefer to be a committer.
> >
> > I would also like to replace the sun jars, just to cut down on the warning
> > messages.
> 
> Okay - here's the deal. Submit a patch that contains the changes to the code 
> of NG4J that are necessary to compile with Jena 2.5.7. I will have a look at 
> them and commit them to our repository. When they are in our codebase I will 
> change your status to Developer so that you can commit in the future.


I have added the patch to the tracker, but also attach here. 

What I have not done is changed any jars in /lib. 
Those changes would be to update antlr to version 2.7.7 
and replace jena.jar and jenatest.jar with jena 2.5.7 jars

The bouncycastle jars should not be different, we are just using them directly fromn the maven repository now. 

I don't know if you are happy to continue to deploy to Melati.org or 
whether there is a desire to get the ng4j jars into the main maven repos. 
It is actually a bit of work to get them in and keep them updated. 
(See the bouncy castle process)

My sourceforge id is timp

cheers
Tim Pizey








 

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Olaf H. <ha...@in...>]

Hey Tim,

On Thursday 23 July 2009 13:04:13 Tim Pizey wrote:
> [...]
> Cool, sorry to be pushy, but the patch creation, submission, review cycle
> is quite a barrier and the more contributions I have the higher my Ohloh
> kudos. https://www.ohloh.net/accounts/timp

Hahaha - I understand ;-)

> In fact what would work better from the point of view of the droids is for
> me to submit the patches, and once you have reviewed them for me to commit
> them, as Ohloh counts individual commits.

I am fine with that.

Olaf

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Tim P. <tim...@ou...>]

On Thursday 23 Jul 2009 11:41:33 Olaf Hartig wrote:
> On Wednesday 22 July 2009 14:35:55 Tim Pizey wrote:
> > > I suspect that there will be a bit of work here, rather than submitting
> > > patches I would prefer to be a committer.
> >
> > I would also like to replace the sun jars, just to cut down on the warning
> > messages.
> 
> Okay - here's the deal. Submit a patch that contains the changes to the code 
> of NG4J that are necessary to compile with Jena 2.5.7. I will have a look at 
> them and commit them to our repository. When they are in our codebase I will 
> change your status to Developer so that you can commit in the future.

Cool, sorry to be pushy, but the patch creation, submission, review cycle is quite  
a barrier and the more contributions I have the higher my Ohloh kudos. 
https://www.ohloh.net/accounts/timp

In fact what would work better from the point of view of the droids is for me to submit the patches, 
and once you have reviewed them for me to commit them, as Ohloh counts individual commits. 

cheers
Tim
goes off to create a patch 

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Olaf H. <ha...@in...>]

Hey Tim,

On Wednesday 22 July 2009 14:35:55 Tim Pizey wrote:
> [...]
> I now have everything building with 2.5.7 and with the BouncyCastle jars
> from central repository.

That's cool.

> > I suspect that there will be a bit of work here, rather than submitting
> > patches I would prefer to be a committer.
>
> I would also like to replace the sun jars, just to cut down on the warning
> messages.

Okay - here's the deal. Submit a patch that contains the changes to the code 
of NG4J that are necessary to compile with Jena 2.5.7. I will have a look at 
them and commit them to our repository. When they are in our codebase I will 
change your status to Developer so that you can commit in the future.

Greetings,
Olaf

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Tim P. <tim...@ou...>]

On Wednesday 22 Jul 2009 12:26:24 Tim Pizey wrote:
> Hi Olaf, 
> 
> I am going to have another run at this. 
> 
> Not the least problem is that we are using a version of Jena which 
> is so old that it is not in the Maven repository (2.5.7 was the first).
>
> I currently have an installation which does not compile with Maven. 


I now have everything building with 2.5.7 and with the BouncyCastle jars from central repository. 
 
> I suspect that there will be a bit of work here, rather than submitting patches I 
> would prefer to be a committer. 


I would also like to replace the sun jars, just to cut down on the warning messages. 

cheers
tim

Re: [namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Tim P. <tim...@ou...>]

On Thursday 09 Jul 2009 19:58:13 Olaf Hartig wrote:
> Hey Tim
> 
> On Thursday 09 July 2009 10:50:13 Tim Pizey wrote:
> > [...]
> > Any progress with moving up to current jena?
> 
> I think you are the only one so far who really tried to upgrade NG4J to the 
> latest Jena release. On June 19 you reported that "most changes are pretty 
> trivial" except one issue with List<NamedGraph>. I assume you have not solved 
> this issue since then. However, could you please send me a patch that 
> contains these other, trivial changes so that I may have a look?


Hi Olaf, 

I am going to have another run at this. 

Not the least problem is that we are using a version of Jena which 
is so old that it is not in the Maven repository (2.5.7 was the first).

I currently have an installation which does not compile with Maven. 

I suspect that there will be a bit of work here, rather than submitting patches I 
would prefer to be a committer. 

yours
Tim

[namedgraphs] Update to latest Jena (Was: Bouncy Castle now in Maven Repository)

[Olaf H. <ha...@in...>]

Hey Tim

On Thursday 09 July 2009 10:50:13 Tim Pizey wrote:
> [...]
> Any progress with moving up to current jena?

I think you are the only one so far who really tried to upgrade NG4J to the 
latest Jena release. On June 19 you reported that "most changes are pretty 
trivial" except one issue with List<NamedGraph>. I assume you have not solved 
this issue since then. However, could you please send me a patch that 
contains these other, trivial changes so that I may have a look?

Greetings,
Olaf

[namedgraphs] Bouncy Castle now in Maven Repository

[Tim P. <tim...@ou...>]

Hi, 

I have just been notified that the Bouncy Castle jars are now in the central Maven repository: 

 http://jira.codehaus.org/browse/MAVENUPLOAD-2479?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel 

Carlos Sanchez closed MAVENUPLOAD-2479.
---------------------------------------

      Assignee: Carlos Sanchez  (was: Brian Fox)
    Resolution: Fixed


I hope to get time to return to this soon. 

Any progress with moving up to current jena?

cheers
TimP

[namedgraphs] Jena2 version 2.5.6 deployed to maven repo

[Tim P. <tim...@ou...>]

Hi, 

Not sure if anyone else is using maven but I have deployed 
the  jar, source and tests for jena2 version 2.5.6  to http://paneris.net/maven2/ 
this enables a you to get the sources code inside an IDE such as eclipse. 
The central maven repository does not contain the source or test jars.

What I did was to checkout to tag Jena2_2-5_6.
Move source files into src/main/java, src/test/java

mv   src/test/java/com/hp/hpl/jena/graph/test/NodeCreateUtils.java src/main/java/com/hp/hpl/jena/graph/test/

Edit src/main/java/com/hp/hpl/jena/ontology/OntTools.java to remove reference to test package. 
Comment out arq tests

Modify pom from current jena2 to revert dependencies to those in lib directory

and deployed. 

hope this helps
Tim

Re: [namedgraphs] Default / unlabelled graphs in TriG

[Olaf H. <ha...@in...>]

On Friday 26 June 2009 12:50:07 Olaf Hartig wrote:
> On Thursday 25 June 2009 15:06:45 Richard Cyganiak wrote:
> > Olaf,
> >
> > Originally TriG did not allow unlabelled graphs, but this was changed
> > after SPARQL was introduced. The test might predate that change. Or
> > not, I'm not sure.
>
> I did. [...]

I meant: *It* did.

Sorry,
Olaf

Re: [namedgraphs] tests/de/fuberlin/wiwiss/ng4j/trig/TriGParserTest.java

[Tim P. <tim...@ou...>]

On Friday 26 Jun 2009 11:52:43 Olaf Hartig wrote:
> Hey Tim,
> 
> On Tuesday 23 June 2009 17:25:05 Olaf Hartig wrote:
> > On Monday 15 June 2009 12:32:26 Tim Pizey wrote:
> > > Hi,
> > >
> > > I have to comment this test out on linux as it has unencoded
> > > high value ascii string literals.
> >
> > What exactly happens with this test enabled on your machine. I am using
> > Linux too and I'm not aware of a problem here.
> 
> I guess I fixed the problem. By comparing the code of this test to a very old 
> NG4J version I found that it was an encoding error that sneaked in.

Yay, good work. 

Tim

Re: [namedgraphs] tests/de/fuberlin/wiwiss/ng4j/trig/TriGParserTest.java

[Olaf H. <ha...@in...>]

Hey Tim,

On Tuesday 23 June 2009 17:25:05 Olaf Hartig wrote:
> On Monday 15 June 2009 12:32:26 Tim Pizey wrote:
> > Hi,
> >
> > I have to comment this test out on linux as it has unencoded
> > high value ascii string literals.
>
> What exactly happens with this test enabled on your machine. I am using
> Linux too and I'm not aware of a problem here.

I guess I fixed the problem. By comparing the code of this test to a very old 
NG4J version I found that it was an encoding error that sneaked in.

Greetings,
Olaf

Re: [namedgraphs] Default / unlabelled graphs in TriG

[Olaf H. <ha...@in...>]

On Thursday 25 June 2009 15:06:45 Richard Cyganiak wrote:
> Olaf,
>
> Originally TriG did not allow unlabelled graphs, but this was changed
> after SPARQL was introduced. The test might predate that change. Or
> not, I'm not sure.

I did. I fixed this problem.

Thanks for clarification.

Olaf

Re: [namedgraphs] N3JenaReader

[Richard C. <ri...@cy...>]

On 23 Jun 2009, at 17:00, Olaf Hartig wrote:
>> One of the quite small number of issues is that N3Reader has been  
>> removed.
>> The CVS Log
>> http://jena.cvs.sourceforge.net/viewvc/jena/jena2/src/com/hp/hpl/jena/n3/N3
>> JenaReader.java?hideattic=0&view=log says:
>> Remove antlr-based N3 parser and route to JavaCC Turtle parser.
>>
>> Is there a drop-in replacement?
>
> Not that I know of :-(
> Anyone else?
> Does this mean we also have to use the JavaCC Turtle parser?

I based the TriG parser on Jena's N3 parser, which was implemented  
using AntLR.

Jena has since removed that parser and replaced it with a Turtle  
parser, which uses JavaCC.

I think you have two options:

a) pull the AntLR bits from the old Jena repository and put them into  
NG4J

b) create a new TriG parser based on Jena's new Turtle parser.

For someone who is already familiar with JavaCC, option b) is likely  
not too much work, and would produce a cleaner and more maintainable  
result. But a) would be quicker and maybe Good Enough.

Best,
Richard



>
> Olaf
>
> ------------------------------------------------------------------------------
> Are you an open source citizen? Join us for the Open Source Bridge  
> conference!
> Portland, OR, June 17-19. Two days of sessions, one day of  
> unconference: $250.
> Need another reason to go? 24-hour hacker lounge. Register today!
> http://ad.doubleclick.net/clk;215844324;13503038;v?http://opensourcebridge.org
> _______________________________________________
> ng4j-namedgraphs mailing list
> ng4...@li...
> https://lists.sourceforge.net/lists/listinfo/ng4j-namedgraphs

Re: [namedgraphs] Default / unlabelled graphs in TriG

[Richard C. <ri...@cy...>]

Olaf,

Originally TriG did not allow unlabelled graphs, but this was changed  
after SPARQL was introduced. The test might predate that change. Or  
not, I'm not sure.

Now you need to check the date when the testUnlabelledGraph test was  
added, and correlate this with the changelog at the end of the TriG  
spec. Hopefully this answers your question ...

Oh, and maybe this helps:
http://dowhatimean.net/2005/10/trig-syntax-issues

Best,
Richard


On 23 Jun 2009, at 17:35, Olaf Hartig wrote:

> Hey Chris and Richard,
>
> (CCing to the NG4J list)
>
> According to  the TriG spec. at
>
> http://www4.wiwiss.fu-berlin.de/bizer/TriG/
>
> TriG allows default graphs which are unlabelled - see example 3 as  
> well as
> the 'graph' rule in the EBNF grammar. Is this correct?
>
> In the JUnit tests for the TriG parser in NG4J we have a test
> case 'testUnlabelledGraph' (line 54 in file
> tests/de/fuberlin/wiwiss/ng4j/trig/TriGParserTest.java) that  
> explicitly
> rejects unlabelled graphs.
>
> I'm puzzled. Are unlabelled default graphs allowed or disallowed in  
> TriG?
>
> Greetings,
> Olaf

[namedgraphs] Default / unlabelled graphs in TriG

[Olaf H. <ha...@in...>]

Hey Chris and Richard,

(CCing to the NG4J list)

According to  the TriG spec. at

 http://www4.wiwiss.fu-berlin.de/bizer/TriG/

TriG allows default graphs which are unlabelled - see example 3 as well as 
the 'graph' rule in the EBNF grammar. Is this correct?

In the JUnit tests for the TriG parser in NG4J we have a test 
case 'testUnlabelledGraph' (line 54 in file 
tests/de/fuberlin/wiwiss/ng4j/trig/TriGParserTest.java) that explicitly 
rejects unlabelled graphs.

I'm puzzled. Are unlabelled default graphs allowed or disallowed in TriG?

Greetings,
Olaf

Re: [namedgraphs] tests/de/fuberlin/wiwiss/ng4j/trig/TriGParserTest.java

[Olaf H. <ha...@in...>]

On Monday 15 June 2009 12:32:26 Tim Pizey wrote:
> Hi,
>
> I have to comment this test out on linux as it has unencoded
> high value ascii string literals.

What exactly happens with this test enabled on your machine. I am using Linux 
too and I'm not aware of a problem here.

Olaf

Re: [namedgraphs] Todo for next version

[Olaf H. <ha...@in...>]

On Friday 19 June 2009 17:53:25 Tim Pizey wrote:
> Hi Olaf, Jennifer and all,
>
> I have started to bringNG4J inline with current Jena.
>
> Most changes are pretty trivial. However there appears to be a
> hack where we treat List<NamedGraph> and List<Graph>
> as covariant which will no longer work.

Where does this "problem" surface - in the constructors of the UnionGraph 
class?
It is strange because a NamedGraph is a Graph. For this reason the compiler 
should understand that a list of NamedGraph objects can be used as a list of 
Graph objects.

> Do you have a plan,

Not really a good one. We could use List<Graph> instead of List<NamedGraph> 
and since we add only NamedGraph objects to this list we can safely cast each 
member of the list to a NamedGraph. However, that's a hack. What do you (and 
the others) think?

> are there other pitfalls? 

I don't know.

Olaf

Re: [namedgraphs] N3JenaReader

[Olaf H. <ha...@in...>]

Hey Tim,

Sorry for the late reply - I was busy finishing a paper.

On Friday 19 June 2009 16:45:36 Tim Pizey wrote:
> Hi,
>
>
> I am trying to upgrade NG4J to the latest Jena.

Great!

> One of the quite small number of issues is that N3Reader has been removed.
> The CVS Log 
> http://jena.cvs.sourceforge.net/viewvc/jena/jena2/src/com/hp/hpl/jena/n3/N3
>JenaReader.java?hideattic=0&view=log says:
> Remove antlr-based N3 parser and route to JavaCC Turtle parser.
>
> Is there a drop-in replacement?

Not that I know of :-(
Anyone else?
Does this mean we also have to use the JavaCC Turtle parser?

Olaf