knobot-devel — Development discussion list

[Knobot-devel] sorry for spam

[Reto Bachmann-G. <re...@gm...>]

now i think I finally found the correct settings to avoid this nuisance
in future.

btw, I hope you noticed the recent changes in knobot, all the jars now
being OSGi bundles and the RDF abstraction layers allowing it to run on
sesame.

Cheers,
reto

[Knobot-devel] Hello

[Chen H. H. L. <Po...@ch...>]

Dear Sir/madam,



I am the GBM (Chairman) of Chen Hsong Holdings Limited   This Company Established since 1958. After half a century of endless efforts, Chen Hsong has grown from a small machinery workshop to one of the largest manufacturers of injection moulding machines in the world.Chen Hsong Holdings Limited, Produces, Exclusive circular platen (patended), ichen shop- floor Networked management system, Ductile iron casting and machining, Jetmaster minijet series, Jetmaster MKIV series, Jetmaster large series, CHEN-PET Two-stage preform moulding turnkey system, jetmaster C Series, E.T.C. Due to long association with our suppliers and our thorough understanding of the working condition in the Industry; It is upon this note that we are writing you this mail to seek your assistance in representing our 

company in your locality as our RECEIVING AGENT/REPRESENTATIVE.One who will act as a medium for our clients in those locality to be reaching us with  their  payments and so on. Note that as a receiving Agent of our company,You will be entitled to TEN Percent Payment of any amount you receive from our customers We seek your Sincere cooperation and assistance to establish a cordial relationship with our clients.To facilitate the conclusion of this proposal if accepted,Please send us 

the  following Information 



1) Your Full name..............and present occupation............



2) Telephone number..............and Fax..............



3) Contact address................



4) Age................



Thanks in anticipation.

Regards,

Dr. Chiang Chen

GBM (Chairman)

Chen Hsong Holdings Limited

[Knobot-devel] Mavenization

[Reto Bachmann-G. <re...@gm...>]

Mosts parts of Knobot are now mavenized in the svn repository (trunk). 
http://gmuer.ch/ is run by this version, as the IEFixFilter has not yet 
been ported (nor manually added to the classpath) it probably doesn't 
work in many browser (it does with ff and opera). The maven repository 
to get the modules is http://wymiwyg.berlios.de/maven2/.

Re: [Knobot-devel] a question on the ordering of TypeHandler

[<re...@gm...>]

dbu wrote:
> hi reto,
>
> thanks for this excplications. now i see better how requests are
> processed.
> the problem with caching is that firefox ignores the header.
> Vary: Cookies
It shouldn't, and I vaguely remember having used successfully somewhere.
Are you referring to a known FF-bug?

reto

Re: [Knobot-devel] a question on the ordering of TypeHandler

[dbu <db...@us...>]

hi reto,

thanks for this excplications. now i see better how requests are processed.
the problem with caching is that firefox ignores the header.
Vary: Cookies
what i am trying to do now is adding a flag to the Response object 
telling whether it may be cached or not. the IEFixFilter could check it 
and only add the no-cache headers if the response is not cachable. 
although it is not the most elegant solution, this would solve the 
biggest problems with caching today: images which do not exist in 
different language variants and the css files would be cached, really 
improving user experience when surfing the page.

i could create a post-filter chain that is used after the 
typechainfilter finished using the handlers and move the IEFixFilter 
there. do you think this is a good idea? or is there a better way?
i could also put a method to the response to remove headers and remove 
all cache-related headers in the fileserverfilter if the response is 
cachable - but this is definitely not the proper way...

david



Reto Bachmann-Gmür wrote:
> hi david,
> 
> - FileServerFilter is a rwcf:TypeHandler
> - IEFixFilter is a rwcf:Filter
> - rwcf:TypeHandlerS get executed by the TypeChainFilter which is an
> rwcf:Filter.
> 
> What happens is the following:
> 1. the rwcf:FilterS gets ordered by priority, higher priority are
> excecuted first, lower if and only if the higher-priority filters
> forward the request
> 2. With prirority 500 the TypeHandlerFilter comes into action, this
> creates a chain of TypeHandler so that what's returned by one
> TypeHandler is the input-type for the next, the goal is to create a
> chain which return a content-type acceptable by the client agent.
> 
> From: http://development.osar.ch/documentation (this should be moved to
> a wymiwyg-site...)
> 
> The Type Handler Filter is a filter which creates a list of
> TypeHandlers  and forwards the request to this list. TypeHanlers are
> itself very similar to Filters, in fact they implement the same java class.
> 
>     * The description of a TypeHandler contains the media type it
> returns, if the handler requires some input data the media-types such
> data must be is specified. (an exception for TypeBlindHandlers as well
> as for TypeHandlers implementing the Java Interface TypeHandler, this
> rarely used features need some redesign)
>     * The desciption of a TypeHandler may force authentification by
> being of type rwcf:PermissionEnforcer
> 
> 
> The Type Hanlder Filter constructs the chain of TypeHandlers  by
> iterating through the available TypeHandlers  in descending order of
> rwcf:priority, when a TypeHandlers matches the current request and  has
> the desired target-typeit if added to the chainand the iteration through
> the available TypeHandlers starts over omiting  those already in the
> chain. This processed is repeated till a terminal chain is found, a
> terminal chain is one that ends with a TypeHandler which needs no input
> data.
> basing on the request it receives so that the changes made by previous
> Filters inluence the selection of TypeHandlers.
> 
> 3. If no type-handler chain can be created the request is forwarded to
> the next rwcf:Filter after TypeHandlerFilter, possibly getting at a
> filter generating a 404 or a method-not-supported error page.
> 
> 
> So setting a low priority to IEFixFilter will cause the filter to be
> executed iff no previous filter can handle the request, which includes
> that no type-handler filter chain can be created, which in turn includes
> that FileServerFilter cannot serve a file.
> 
> reto
> 
> PS: http://jena.hpl.hp.com/juc2006/proceedings/bachmann-gmur/paper.pdf
> also has some infos about FilterS and TypeHandlerS
> 
> dbu wrote:
>> hi,
>>
>> i am experimenting with caching and the IEFixFilter.
>> i realised that the FileServerFilter gets called after the IEFixFilter. 
>> the iefix-section.rdf defines the priority of the iefix, fs-section.rdf 
>> that of the fileserver.
>> i tried to put iefixfilter priority lower than that of the filefilter, 
>> but this results in iefixfilter not being used at all.
>> how can i control which filters are used on a request and in what order? 
>> or can't i use several handlers on one request?
>>
>> cheers,david
>>
>> -------------------------------------------------------------------------
>> This SF.net email is sponsored by DB2 Express
>> Download DB2 Express C - the FREE version of DB2 express and take
>> control of your XML. No limits. Just data. Click to get it now.
>> http://sourceforge.net/powerbar/db2/
>> _______________________________________________
>> Knobot-devel mailing list
>> Kno...@li...
>> https://lists.sourceforge.net/lists/listinfo/knobot-devel
>>   
> 
> 

Re: [Knobot-devel] a question on the ordering of TypeHandler

[<re...@gm...>]

hi david,

- FileServerFilter is a rwcf:TypeHandler
- IEFixFilter is a rwcf:Filter
- rwcf:TypeHandlerS get executed by the TypeChainFilter which is an
rwcf:Filter.

What happens is the following:
1. the rwcf:FilterS gets ordered by priority, higher priority are
excecuted first, lower if and only if the higher-priority filters
forward the request
2. With prirority 500 the TypeHandlerFilter comes into action, this
creates a chain of TypeHandler so that what's returned by one
TypeHandler is the input-type for the next, the goal is to create a
chain which return a content-type acceptable by the client agent.

From: http://development.osar.ch/documentation (this should be moved to
a wymiwyg-site...)

The Type Handler Filter is a filter which creates a list of
TypeHandlers  and forwards the request to this list. TypeHanlers are
itself very similar to Filters, in fact they implement the same java clas=
s.

    * The description of a TypeHandler contains the media type it
returns, if the handler requires some input data the media-types such
data must be is specified. (an exception for TypeBlindHandlers as well
as for TypeHandlers implementing the Java Interface TypeHandler, this
rarely used features need some redesign)
    * The desciption of a TypeHandler may force authentification by
being of type rwcf:PermissionEnforcer


The Type Hanlder Filter constructs the chain of TypeHandlers  by
iterating through the available TypeHandlers  in descending order of
rwcf:priority, when a TypeHandlers matches the current request and  has
the desired target-typeit if added to the chainand the iteration through
the available TypeHandlers starts over omiting  those already in the
chain. This processed is repeated till a terminal chain is found, a
terminal chain is one that ends with a TypeHandler which needs no input
data.
basing on the request it receives so that the changes made by previous
Filters inluence the selection of TypeHandlers.

3. If no type-handler chain can be created the request is forwarded to
the next rwcf:Filter after TypeHandlerFilter, possibly getting at a
filter generating a 404 or a method-not-supported error page.


So setting a low priority to IEFixFilter will cause the filter to be
executed iff no previous filter can handle the request, which includes
that no type-handler filter chain can be created, which in turn includes
that FileServerFilter cannot serve a file.

reto

PS: http://jena.hpl.hp.com/juc2006/proceedings/bachmann-gmur/paper.pdf
also has some infos about FilterS and TypeHandlerS

dbu wrote:
> hi,
>
> i am experimenting with caching and the IEFixFilter.
> i realised that the FileServerFilter gets called after the IEFixFilter.=
=20
> the iefix-section.rdf defines the priority of the iefix, fs-section.rdf=
=20
> that of the fileserver.
> i tried to put iefixfilter priority lower than that of the filefilter, =

> but this results in iefixfilter not being used at all.
> how can i control which filters are used on a request and in what order=
?=20
> or can't i use several handlers on one request?
>
> cheers,david
>
> -----------------------------------------------------------------------=
--
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Knobot-devel mailing list
> Kno...@li...
> https://lists.sourceforge.net/lists/listinfo/knobot-devel
>  =20

[Knobot-devel] a question on the ordering of TypeHandler

[dbu <db...@us...>]

hi,

i am experimenting with caching and the IEFixFilter.
i realised that the FileServerFilter gets called after the IEFixFilter. 
the iefix-section.rdf defines the priority of the iefix, fs-section.rdf 
that of the fileserver.
i tried to put iefixfilter priority lower than that of the filefilter, 
but this results in iefixfilter not being used at all.
how can i control which filters are used on a request and in what order? 
or can't i use several handlers on one request?

cheers,david

[Knobot-devel] solved: add relation loops forever: empty title

[David B. <db...@us...>]

hi reto,

thanks, that solved it. i reply to the mailinglist so maybe others can 
find this message if they run into the same problem.

i needed to link one topic into another (Add Relation). i pasted that 
relation and clicked on Continue, but nothing happened (except the 
spinning circle turning forever)

reto suggested that this might be a javascript problem. RDFModel.js 
could not handle that the title of the target was empty (at least in my 
current language). after i put a title to the target topic in all 
languages, linking worked fine.

thanks reto!
cu david

Re: [Knobot-devel] next-relation, permission changes and restarts

[<re...@gm...>]

hi,

may I suggest you download the dump without tempory triples immediately
before and after a restart to see more precisely what gets lots.

do you have the same problem when using the same data on a test
installation?

reto

David Buchmann wrote:
> hi reto,
>
> the system is able to write some things. articles and topics do not get=
=20
> lost on restart, they stay in the database. i checked and the disk has =

> more than 1 GB free. if articles are stored using rdbloader, the=20
> database is still accepting and storing data.
>
>
> [ISO-8859-1] Reto Bachmann-Gm?r writes:
>
>  > hi,
>  >
>  > this is strange indeed.
>  >
>  > one possible explanation: for some reason the system fails to write
>  > changes (disk full, postgeress problems) the changes can however be
>  > written to the in-memory cache and diasappear only after a restart. =
Do
>  > regular changes (like adding articles) survive restarts?
>  >
>  > reto
>  >
>  > dbu wrote:
>  >> hi,
>  >>
>  >> i have a problem with our knobot installation. after restarting, it=

>  >> forgets all newly added next-relation/first-relation statements. th=
is
>  >> can be fixed by first "Remove next and first-relation stmts"
>  >> (orphanize-all) and then "De-Orphanize Relations" (de-orphanize) - =
which
>  >> is however an awkward solution and takes ~5 minutes on our site.
>  >> for the permission, we have at least one article where we know of t=
he
>  >> problem: it was first created with requiring 'Edit' permission. lat=
er
>  >> the permission was removed, but after each restart, the permission
>  >> reappears. we did not find a way to fix this problem.
>  >>
>  >> we use rdbloader with postgres.
>  >> i dumped the whole model as n-triples, deleted the complete postgre=
s
>  >> database and then restarted knobot. on startup, it re-initialized t=
he
>  >> database and i uploaded the model. however, this does not fix the
>  >> problem: whatever we create after the restore looses its relations.=
 (the
>  >> articles / topics are still in the database, though)
>  >>
>  >> any ideas?
>  >>
>  >> kind regards,david
>
>
> -----------------------------------------------------------------------=
--
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Knobot-devel mailing list
> Kno...@li...
> https://lists.sourceforge.net/lists/listinfo/knobot-devel
>  =20

Re: [Knobot-devel] next-relation, permission changes and restarts

[David B. <db...@us...>]

hi reto,

the system is able to write some things. articles and topics do not get 
lost on restart, they stay in the database. i checked and the disk has 
more than 1 GB free. if articles are stored using rdbloader, the 
database is still accepting and storing data.


[ISO-8859-1] Reto Bachmann-Gm?r writes:

 > hi,
 >
 > this is strange indeed.
 >
 > one possible explanation: for some reason the system fails to write
 > changes (disk full, postgeress problems) the changes can however be
 > written to the in-memory cache and diasappear only after a restart. Do
 > regular changes (like adding articles) survive restarts?
 >
 > reto
 >
 > dbu wrote:
 >> hi,
 >>
 >> i have a problem with our knobot installation. after restarting, it
 >> forgets all newly added next-relation/first-relation statements. this
 >> can be fixed by first "Remove next and first-relation stmts"
 >> (orphanize-all) and then "De-Orphanize Relations" (de-orphanize) - which
 >> is however an awkward solution and takes ~5 minutes on our site.
 >> for the permission, we have at least one article where we know of the
 >> problem: it was first created with requiring 'Edit' permission. later
 >> the permission was removed, but after each restart, the permission
 >> reappears. we did not find a way to fix this problem.
 >>
 >> we use rdbloader with postgres.
 >> i dumped the whole model as n-triples, deleted the complete postgres
 >> database and then restarted knobot. on startup, it re-initialized the
 >> database and i uploaded the model. however, this does not fix the
 >> problem: whatever we create after the restore looses its relations. (the
 >> articles / topics are still in the database, though)
 >>
 >> any ideas?
 >>
 >> kind regards,david

Re: [Knobot-devel] next-relation, permission changes and restarts

[<re...@gm...>]

hi,

this is strange indeed.

one possible explanation: for some reason the system fails to write
changes (disk full, postgeress problems) the changes can however be
written to the in-memory cache and diasappear only after a restart. Do
regular changes (like adding articles) survive restarts?

reto

dbu wrote:
> hi,
>
> i have a problem with our knobot installation. after restarting, it
> forgets all newly added next-relation/first-relation statements. this
> can be fixed by first "Remove next and first-relation stmts"
> (orphanize-all) and then "De-Orphanize Relations" (de-orphanize) - whic=
h
> is however an awkward solution and takes ~5 minutes on our site.
> for the permission, we have at least one article where we know of the
> problem: it was first created with requiring 'Edit' permission. later
> the permission was removed, but after each restart, the permission
> reappears. we did not find a way to fix this problem.
>
> we use rdbloader with postgres.
> i dumped the whole model as n-triples, deleted the complete postgres
> database and then restarted knobot. on startup, it re-initialized the
> database and i uploaded the model. however, this does not fix the
> problem: whatever we create after the restore looses its relations. (th=
e
> articles / topics are still in the database, though)
>
> any ideas?
>
> kind regards,david
>
>
> -----------------------------------------------------------------------=
--
> This SF.net email is sponsored by DB2 Express
> Download DB2 Express C - the FREE version of DB2 express and take
> control of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> Knobot-devel mailing list
> Kno...@li...
> https://lists.sourceforge.net/lists/listinfo/knobot-devel
>  =20

[Knobot-devel] next-relation, permission changes and restarts

[dbu <db...@us...>]

hi,

i have a problem with our knobot installation. after restarting, it
forgets all newly added next-relation/first-relation statements. this
can be fixed by first "Remove next and first-relation stmts"
(orphanize-all) and then "De-Orphanize Relations" (de-orphanize) - which
is however an awkward solution and takes ~5 minutes on our site.
for the permission, we have at least one article where we know of the
problem: it was first created with requiring 'Edit' permission. later
the permission was removed, but after each restart, the permission
reappears. we did not find a way to fix this problem.

we use rdbloader with postgres.
i dumped the whole model as n-triples, deleted the complete postgres
database and then restarted knobot. on startup, it re-initialized the
database and i uploaded the model. however, this does not fix the
problem: whatever we create after the restore looses its relations. (the
articles / topics are still in the database, though)

any ideas?

kind regards,david

[Knobot-devel] error handling

[<re...@gm...>]

david pointed out some problems with the error handling in KnoBot:

Situation:
- If an exception occurs before the body has been committed one get's an
errorpage, showing both the error message and the stack-trace
- If an exception occurs after the body has been committed a Jetty error
page for the error "response already committed" shows up
- If an exception in the transforming thread occurs, a white page is
returned

The first situation is what actually should happen, the template can be
changed not to display the stack-trace. The latest situation should
probably result n the exception being forwarded to the handler running
the thread, which would usually boil it down to the problematic
situation two: how to handle error when the response has already been
committed. The answer might be: do it as jetty does, not sure how though.=


Cheers,
reto

[Knobot-devel] xhtml

[<re...@gm...>]

As discossuded at the Knobot meeting, I changed the iefix-filter no
longer to change the xhtml content-type to html if the client
explicitely accepts xhtml. as a consequence the definition of those
templates which require the result to be interpreted as html (i.e. those
using tinymce) need to be redefined.

Re: [Knobot-devel] knobot failsaveness

[<re...@gm...>]

dbu wrote:
> in order to improve user experience in case of failures, it would be=20
> nice if an error page could be displayed.
> when an error occurs, at the moment all visitors to the page just get a=
=20
> blank page. i guess this would have to happen in=20
> org.wymiwyg.rwcf.tools.xslt.XSLTResponseWrapper at line 224?
>
> any objections if i do this?
>  =20
as long as I don't have tto write a documentation for the RWCF exception
handling mechanisms, go ahead!

from rwcf/src/META_INF/rwcf/ModelSection.rdf:
<rwcf:ExceptionHandler
rwcf:javaClass=3D"org.wymiwyg.rwcf.impl.exceptionhandlers.HandlerExceptio=
nHandler"

        rwcf:exceptionClass=3D"java.lang.Exception">
    </rwcf:ExceptionHandler>


from rwcf-tools/notfound/src/META-INF/rwcf/nf-section.rdf:
<rwcf:ExceptionHandler
rwcf:javaClass=3D"org.wymiwyg.rwcf.tools.notfound.NotFoundExceptionHandle=
r"
      =20
rwcf:exceptionClass=3D"org.wymiwyg.rwcf.tools.notfound.NotFoundException"=
>
    </rwcf:ExceptionHandler>

Re: [Knobot-devel] stack overflow

[<re...@gm...>]

hi david,

since version 0.2.7 knobot comes with a custom build of xalan patched to
avoid a memory-leak situation with the symptom you describe. Starting
KnoBot the normal way, set's the endorsed-lib system property so that
the shipped version of xalan is used. Since august 2005 however, a new
version of xalan might have been released, so it might be worth checking
it out.

cheers,
reto

dbu wrote:
> hello everybody,
>
> with really long documents, i get a stack overflow like the one below.
>
> it does not matter whether there are special characters in it or=20
> anything. i just wrote "bla bla " and copied it over and over... the=20
> text is now 138'750 times bla with 1'143 newlines :-)
>
> please do not tell me this is exagerated. we had a document to publish =

> with 45'000 characters, which was enough to run into the error.
>
> any idea why this should result into a stack overflow? i found
> http://xml.apache.org/xalan-j/faq.html#faq-N102CD (increasing -Xss1m di=
d=20
> not help) talking about "deeply recursive stylesheet" and now i wonder =

> what xslt script the response wrapper actually calls? i mean, there=20
> should be no need for that much recursion in a simple text with no=20
> nested elements.
> or is there no recursive stylesheet, and this is a bug of xalan?
>
> kind regards,david
>
> @reto: the page where i played is:=20
> http://development.osar.ch/newsletter-test
>
>
>
> ERROR [Thread-4082] (XSLTResponseWrapper.java:224) -=20
> java.lang.StackOverflowError
> java.lang.StackOverflowError
>          at=20
> org.apache.xml.dtm.ref.ExpandedNameTable.getExpandedTypeID(ExpandedName=
Table.java:185)
>          at=20
> org.apache.xml.dtm.ref.ExpandedNameTable.getExpandedTypeID(ExpandedName=
Table.java:156)
>          at=20
> org.apache.xml.dtm.ref.sax2dtm.SAX2DTM.processingInstruction(SAX2DTM.ja=
va:2133)
>          at=20
> org.apache.xml.serializer.ToXMLSAXHandler.processingInstruction(ToXMLSA=
XHandler.java:582)
>          at=20
> org.apache.xalan.templates.ElemValueOf.execute(ElemValueOf.java:255)
>          at=20
> org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(Tran=
sformerImpl.java:2411)
>          at=20
> org.apache.xalan.templates.ElemChoose.execute(ElemChoose.java:126)
>          at=20
> org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(Tran=
sformerImpl.java:2411)
>          at=20
> org.apache.xalan.templates.ElemTemplate.execute(ElemTemplate.java:392)
>          at=20
> org.apache.xalan.templates.ElemCallTemplate.execute(ElemCallTemplate.ja=
va:246)
>          at=20
> org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(Tran=
sformerImpl.java:2411)
>          at=20
> org.apache.xalan.templates.ElemChoose.execute(ElemChoose.java:126)
>          at=20
> org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(Tran=
sformerImpl.java:2411)
>          at=20
> org.apache.xalan.templates.ElemTemplate.execute(ElemTemplate.java:392)
> [repeated over and over]
>
>
> -----------------------------------------------------------------------=
--
> 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
> _______________________________________________
> Knobot-devel mailing list
> Kno...@li...
> https://lists.sourceforge.net/lists/listinfo/knobot-devel
>  =20

[Knobot-devel] knobot failsaveness

[dbu <db...@us...>]

in order to improve user experience in case of failures, it would be 
nice if an error page could be displayed.
when an error occurs, at the moment all visitors to the page just get a 
blank page. i guess this would have to happen in 
org.wymiwyg.rwcf.tools.xslt.XSLTResponseWrapper at line 224?

any objections if i do this?

[Knobot-devel] stack overflow

[dbu <db...@us...>]

hello everybody,

with really long documents, i get a stack overflow like the one below.

it does not matter whether there are special characters in it or 
anything. i just wrote "bla bla " and copied it over and over... the 
text is now 138'750 times bla with 1'143 newlines :-)

please do not tell me this is exagerated. we had a document to publish 
with 45'000 characters, which was enough to run into the error.

any idea why this should result into a stack overflow? i found
http://xml.apache.org/xalan-j/faq.html#faq-N102CD (increasing -Xss1m did 
not help) talking about "deeply recursive stylesheet" and now i wonder 
what xslt script the response wrapper actually calls? i mean, there 
should be no need for that much recursion in a simple text with no 
nested elements.
or is there no recursive stylesheet, and this is a bug of xalan?

kind regards,david

@reto: the page where i played is: 
http://development.osar.ch/newsletter-test



ERROR [Thread-4082] (XSLTResponseWrapper.java:224) - 
java.lang.StackOverflowError
java.lang.StackOverflowError
         at 
org.apache.xml.dtm.ref.ExpandedNameTable.getExpandedTypeID(ExpandedNameTable.java:185)
         at 
org.apache.xml.dtm.ref.ExpandedNameTable.getExpandedTypeID(ExpandedNameTable.java:156)
         at 
org.apache.xml.dtm.ref.sax2dtm.SAX2DTM.processingInstruction(SAX2DTM.java:2133)
         at 
org.apache.xml.serializer.ToXMLSAXHandler.processingInstruction(ToXMLSAXHandler.java:582)
         at 
org.apache.xalan.templates.ElemValueOf.execute(ElemValueOf.java:255)
         at 
org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(TransformerImpl.java:2411)
         at 
org.apache.xalan.templates.ElemChoose.execute(ElemChoose.java:126)
         at 
org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(TransformerImpl.java:2411)
         at 
org.apache.xalan.templates.ElemTemplate.execute(ElemTemplate.java:392)
         at 
org.apache.xalan.templates.ElemCallTemplate.execute(ElemCallTemplate.java:246)
         at 
org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(TransformerImpl.java:2411)
         at 
org.apache.xalan.templates.ElemChoose.execute(ElemChoose.java:126)
         at 
org.apache.xalan.transformer.TransformerImpl.executeChildTemplates(TransformerImpl.java:2411)
         at 
org.apache.xalan.templates.ElemTemplate.execute(ElemTemplate.java:392)
[repeated over and over]

[Knobot-devel] GVS as KnoBot backup

[<re...@gm...>]

Forwarding my announcement of GVS on the Jena mailing list.

GVS could be used as a way to do frequent backups of a KnoBot store
without wasting disk space. A usage szenario would be a non-public
gvs-store to and a cron-job which uses curl to get the model from knobot
and put it to gvs. gvs's aggregation service could be used as well, but
would probably have to be enhanced to support password authentication.

reto

Re: [Knobot-devel] backup / restore

[<re...@gm...>]

dbu wrote:
> if a got it correctly, the proper way to backup a web site driven by=20
> knobot is to download a dump of the model from /admin/dump.nt
>  =20
apart from this you should backup the directory called "hashstore"
> however, when it comes to restore, i have a few questions:
> /admin/upload
>
> - is it possible to overwrite an existing site? or do i have to wipe it=
=20
> clean first?
>  =20
just check "Replace non-temporary statements"
> - when i experimented with a fresh install, i had to first create a use=
r=20
> and knobot initialyzed the database. after uploading the n-triples, all=
=20
> entries in the admin page got duplicated. and i had an italian language=
=20
> choice, where there was only en/fr/de before.
>  =20
check "Remove temporary statement in insertion", this should solve the
duplicated items, not the language problem. the four languages gets
added when knobot found no language at all (=3D fresh install), the
autocreated languages should be removed when using "Replace
non-temporary statements"
> some more questions about the upload interface:
> - what is the field "Base" for, how is it used?
>  =20
for rdf/xml uploads using relative URIs
> - what should i choose about the "temporary statements"?
>  =20
the temporary statements are the configuration statements provided by
the modules in META-INF/rwcf of the jars, these statements are added
whenever the system boots. If such a triple is deleted it is not
re-inserted, to "undo" all manual deletions use the admin entry "Reset
manully deleted temporyry statements (the system statements will be
reset at next boot)" <http://www.osar.ch/admin/reset-manually-deleted>.

Replace non-temporary statements -> remove all existing triples but the
configuration triples provided by the modules
Remove temporary statement in insertion -> remove configuration triples
from what is uploaded, this is not needed when the backup was made from
/admin/dump.nt?non_temporary=3Dtrue otherwise needed.
> - is it correct that, to have a complete backup, i need the triples and=
=20
> the hashstore? or is there anything else?
well, you're site may be using the old file-server-filter in which case
this files have to be backuped as well, this is not the case if you
upload the files using webdav or http-put in which case the date is in
the model and hashstore.
>  what about imgupload and ftpupload?
>  =20
this has nothing todo with KnoBot, they are just regular folders for
plain-old file-serving on osar.ch (they contain stuff from the old
website, because there are many links around to those old documents).

reto

Re: [Knobot-devel] knobot orphans

[<re...@gm...>]

dbu wrote:
> hello reto,
>
> [taking a direct discussion onto this mailing list]
>
> thanks for your tip how to solve the NullpointerException we got becaus=
e
> one statement had a nextrelation pointing to an non-existent statement.=

>  =20
strange use of "statement" bot are resources, the first of type
knoBot:Relation the second most likely an rss:item
> doing "remove  next and first relation stmts" and then "de-orphanize
> relations" solved it.
>
> i am however afraid of loosing ordering information or something.=20
the statements deleted and recreated are a faster way to access the
ordered relations than getting all relations with a source and sorting
them. if the orer of items change there was something strange before,
such a "strange" situation may happen when the relations have different
aging-values.
> is
> there an easy way how to restrict the OrphanizeAll to only statements
> having a specific source? can you give me a hint how i could implement =
this?
> http://knobot.cvs.sourceforge.net/knobot/knobot/src/org/wymiwyg/knobot/=
admin/OrphanizeAll.java?revision=3D1.3&view=3Dmarkup
> looks like the place to edit, but how?
>  =20
seems unncessary to me, you would have to iterate through resources of
type relation (listSubjectsWithProperty(RDF.type, KNOBOT.Relation),
check if the source matches and delete next/first if applicable.

	=09

reto

[Knobot-devel] backup / restore

[dbu <db...@us...>]

if a got it correctly, the proper way to backup a web site driven by 
knobot is to download a dump of the model from /admin/dump.nt

however, when it comes to restore, i have a few questions:
/admin/upload

- is it possible to overwrite an existing site? or do i have to wipe it 
clean first?
- when i experimented with a fresh install, i had to first create a user 
and knobot initialyzed the database. after uploading the n-triples, all 
entries in the admin page got duplicated. and i had an italian language 
choice, where there was only en/fr/de before.

some more questions about the upload interface:
- what is the field "Base" for, how is it used?
- what should i choose about the "temporary statements"?
- is it correct that, to have a complete backup, i need the triples and 
the hashstore? or is there anything else? what about imgupload and 
ftpupload?


many questions, i hope somebody has some answers :-)
david

[Knobot-devel] knobot orphans

[dbu <db...@us...>]

hello reto,

[taking a direct discussion onto this mailing list]

thanks for your tip how to solve the NullpointerException we got because
one statement had a nextrelation pointing to an non-existent statement.
doing "remove  next and first relation stmts" and then "de-orphanize
relations" solved it.

i am however afraid of loosing ordering information or something. is
there an easy way how to restrict the OrphanizeAll to only statements
having a specific source? can you give me a hint how i could implement this?
http://knobot.cvs.sourceforge.net/knobot/knobot/src/org/wymiwyg/knobot/admin/OrphanizeAll.java?revision=1.3&view=markup
looks like the place to edit, but how?

thx,
david

[Knobot-devel] TODO: caching / headers

[<re...@gm...>]

As Christian Biesinger <mailto:cbi...@gm...> commented, KnoBot
should deliver a "Vary: Cookie" header and the it might be possible to
allow caching.

[Knobot-devel] major commit

[<re...@gm...>]

I'm currently committing changes related to the extraction of
wymiwyg-commons to http://developer.berlios.de/projects/wymiwyg. The
Knobot related projects do not (currently) include the jars of
wymiwyg-commons, so that the two svn repositories must be checked out
for the knobot-related project to compile.the run-target "knobot" has
been changed for the new-classpath, some other run-targets might still
be lying around.

I removed all jars which do not prevent knbot from compiling, starting
and passing the tests. Sadly, the latest criterion beingas week as the
test suite. If something isn't working, i promise to fix it very quickly
if you add a test showing the bug.

Projects depending on wymiwyg-commons are generally fixed by (in eclipse)=
:
- making sure the two new wymiwyg-commons projects are select as
required project (this is the case anyway for projects depending on rwcf)=

- organize imports on the project


reto