|
From: Andrea S. <and...@bb...> - 2011-01-19 12:59:32
|
Hi,
Yes, a real logger would be nice.
I have to restart my laptop to use WireShark... I usually restart it every 2-3 months. So... give me a couple of days! (I do long term ultra multi-tasking...)
ciao,
Andrea
Il giorno 18/gen/2011, alle ore 15.35, Eric Prud'hommeaux ha scritto:
> * Andrea Splendiani <and...@bb...> [2011-01-18 01:15+0000]
>> Hi,
>>
>> I have some problem with SWObjects.
>>
>> first: with debug on it prints a lot of information, but not all so
>> informative. It prints a lot of html (possibly the one that is served) but I
>> don't easily find the query translation (which is pretty useful).
>> Also, I think it prints the binary png all the time the browser request
>> it... which is a bit annoying.
>
> It looks like it's time to put in a real logger with debug levels and that sort of thing. There's a package proposed for the boost libraries; I've been hoping it would be incorporated into the mainstream before I needed it, but I can get around this by incorporating the source directly into the package. Due to lots of demands, this will take me a couple days.
>
> You can reduce the noise but invoking the query you want to investigate from the command line, à la: /Applications/Sparql.app/Contents/MacOS/Sparql --debug 1 -m YourMapFile -npe 'SomeQuery'
>
>
>> I'm trying to put SWObjects in the middle of RelFinder on one side and
>> Joseki on the other. SWObjects starts with a mapping rule which has a
>> Service clause which match to the Joseki endpoint.
>>
>> So far I had mixed experiences.
>>
>> Re: RelFinder.
>> I've used reverse proxy to get around the crosdmain.xml issue.
>> I didn't manage to have RelFinder working on the SWObjects endpoint yet
>> (still have to play a bit with configurations). However SWObjects easily
>> gets to consume all CPU for hours without any debug message. Maybe it gets
>> in some loop ?
>
> Yeah, sounds like an issue with the HTTP server (which needs a lot of hardening). I'm guessing we'll see better reliability if I just do the Apache module hack (as writing a real HTTP 1.1 server is a sizable project). There's also the possibility of using JNI (I've never compiled it on a mac, but is hopefully not too hard) or calling the sparql executable as a CGI script.
>
>
>> Re: Joseki.
>>
>> I had mixed results.
>>
>> Sometimes it works.
>>
>> For a query like:
>>
>> select ?x ?y
>> where {?x <http://www.test.org/causativeOf> ?y }
>>
>> (the predicate is constructed in a mapping file, and that's the only rule).
>>
>>
>>
>> With this query:
>>
>> select ?x ?xl ?y
>> where {?x <http://www.test.org/causativeOf> ?y .
>> ?x ?p1 ?xl }
>>
>> JOSEKI gives this:
>>
>> 01:10:43 INFO SPARQL :: OK/select: SELECT DISTINCT
>> ?_0x10100e800_0_specGene ?x ?_0x10100e800_0_pert ?_0x10100e800_0_interaction
>> ?_0x10100e800_0_phenotype ?y WHERE { ?_0x10100e800_0_specGene
>> <http://localhost:8180/phibase/hasReferenceGene> ?x . ?_0x10100e800_0_pert
>> <http://localhost:8180/phibase/relativeToExperimentSpecificEntity>
>> ?_0x10100e800_0_specGene . ?_0x10100e800_0_interaction
>> <http://localhost:8180/phibase/Perturbation2Interaction>
>> ?_0x10100e800_0_pert . ?_0x10100e800_0_phenotype
>> <http://localhost:8180/phibase/GenericPhenotype_interactionID>
>> ?_0x10100e800_0_interaction . ?_0x10100e800_0_phenotype
>> <http://localhost:8180/phibase/GenericPhenotype_CV-ProcessPhenotype> ?y . }
>> 01:10:47 WARN log :: handle failed
>> java.io.IOException: FULL head
>> at org.mortbay.jetty.HttpParser.parseNext(HttpParser.java:276)
>> at org.mortbay.jetty.HttpParser.parseAvailable(HttpParser.java:205)
>> at org.mortbay.jetty.HttpConnection.handle(HttpConnection.java:380)
>> at
>> org.mortbay.jetty.nio.BlockingChannelConnector$Connection.run(BlockingChannelConnector.java:163)
>> at
>> org.mortbay.thread.QueuedThreadPool$PoolThread.run(QueuedThreadPool.java:488)
>
> Let's see if I understand this failure; JOSEKI is crashing when reading the query, presumably 'cause Sparql is sending a broken HTTP packet? Can you run a packet sniffer like http://www.wireshark.org/download.html ? That should help us see exactly what's going wrong.
>
>
>> Note that the translated query works directly on JOSEKI. It is possibly
>> something that has to do with the http messanging.
>>
>>
>> Sometimes (not easily reproducible) SWOBjects yield nothing but the query
>> doesn't even go to JOSEKI (but this is a bit vague)
>>
>>
>> Any hint ?
>>
>> Regarding deployment. In the long run, it would be nice to have one (or
>> more) SPARQL module(s) in apache conf. But that's not something so urgent.
>
> ... unless it's the shortest path to satisfaction on the above HTTP issues...
>
>
>> ciao,
>> Andrea
>>
>>
>> Il giorno 16/gen/2011, alle ore 17.59, Eric Prud'hommeaux ha scritto:
>>
>>> * Andrea Splendiani <and...@bb...> [2011-01-16
>> 14:45+0000]
>>>> Hi,
>>>>
>>>> now it works, though at times I had some strange error where in the logs
>> you see some binary blurp (like an image). But that's not so common and
>> reproducible, may be something else around the system.
>>>
>>> Is that only when you have --debug enabled?
>>> If so, it's probably your web browser asking for the favicon and the
>>> server responding with a silly "mr yuck" face.
>>>
>>>
>>>> a question: is there a way to serve some file through the same http
>> server which serves sparql ?
>>>> I would need a crossdomain.xml file for flash!
>>>
>>> It would be pretty trivial to hack the source to do that (I could help
>> there).
>>> I'd like to add this functionality in general, but I haven't worked out
>> the packaging issues. Specifically, I'd like to take the favicon and the
>> generated HTML at the root out and put them into the filesystem somewhere.
>> Issues:
>>> where's that somewhere and how does the executable find it? (it should
>> have a convenient zero-config behavior)
>>> what's the interaction between named graphs and the filesystem? GET and
>> PUT of some named graph has nice behaviors now; should they be shadowed in
>> the filesystem?
>>>
>>> TimBL's leaning on me for the filesystem shadow. I have to write some
>> serializers to make it work.
>>>
>>>
>>>> ciao,
>>>> Andrea
>>>>
>>>> Il giorno 16/gen/2011, alle ore 04.42, Eric Prud'hommeaux ha scritto:
>>>>
>>>>> * Andrea Splendiani <and...@bb...> [2011-01-15
>> 17:30+0000]
>>>>>> Hi,
>>>>>>
>>>>>> I was trying to use the SWObjects that you (tele)presented at the
>> SWAT4LS tutorial in December.
>>>>>>
>>>>>> I have a few problems, perhaps I'm getting something wrong:
>>>>>>
>>>>>> So, I'm starting a Sparql server with the following command:
>>>>>>
>>>>>> ./Sparql --debug 2 -npm
>> /Users/andreasplendiani/Dropbox/WorkingProjects/PhiBase2/phibaseRDFMap.mp
>> --serve http://localhost:8200/SPARQL -i
>> /Users/andreasplendiani/Dropbox/WorkingProjects/PhiBase2/d2r-server-0.7/PhiBaseRDFv003.n3
>>
>>>>>>
>>>>>> The maps is something as:
>>>>>>
>>>>>> CONSTRUCT
>>>>>> {
>>>>>> ?gene <http://www.test.org/causativeOf> ?specGene
>>>>>> }
>>>>>> WHERE {
>>>>>> ?specGene <http://localhost:8180/phibase/hasReferenceGene> ?gene
>>>>>> }
>>>>>>
>>>>>> When I query the endpoint, I get:
>>>>>>
>>>>>> (endpoint result):
>>>>>>
>>>>>> Q&D SPARQL Server Error
>>>>>>
>>>>>> SELECT ?x ?y ?z
>>>>>> where {?x ?y ?z}
>>>>>>
>>>>>> yeilded
>>>>>>
>>>>>> failed to match triples prefixed by "!" in
>>>>>> SELECT ?x ?y ?z
>>>>>> WHERE
>>>>>> {
>>>>>> ! ?x ?y ?z .
>>>>>> }
>>>>>
>>>>> that version didn't support wildcard predicates over mapped rules. could
>> you try the latest?
>>>>>
>>>>>> (log):
>>>>>>
>>>>>> 127.0.0.1 - - [2011-00-15T17:23:00]"GET /favicon.ico 1.1" 200 620
>>>>>> GET /favicon.ico HTTP/1.1
>>>>>> Host: 127.0.0.1:8200
>>>>>> User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; en-US;
>> rv:1.9.2.13) Gecko/20101203 Firefox/3.6.13
>>>>>> Accept:
>> text/html,application/xhtml+xml,application/xml;q=0.9,*/*;q=0.8
>>>>>> Accept-Language: en-us,en;q=0.5
>>>>>> Accept-Encoding: gzip,deflate
>>>>>> Accept-Charset: ISO-8859-1,utf-8;q=0.7,*;q=0.7
>>>>>> Keep-Alive: 115
>>>>>> Connection: keep-alive
>>>>>> Cookie:
>> __utma=96992031.761815913.1280272954.1288624527.1291566200.4;
>> __utmz=96992031.1280272954.1.1.utmcsr=(direct)|utmccn=(direct)|utmcmd=(none)
>>>>>>
>>>>>>
>>>>>> HTTP/1.0 200 OK
>>>>>> Content-Type: image/x-icon
>>>>>>
>>>>>> ?PNG
>>>>>>
>>>>>> (then binary blurp).
>>>>>>
>>>>>> Not sure what's going on, any idea ?
>>>>>> Same things if I try via SERVICE (instead of actually pointing to a
>> local dataset, which I suspect is not read)
>>>>>>
>>>>>> Is there a mailing list for SWObjects ?
>>>>>>
>>>>>> ciao,
>>>>>> Andrea
>>>>>>
>>>>>>
>>>>>>
>>>>>> Andrea Splendiani
>>>>>> Senior Bioinformatics Scientist
>>>>>> Centre for Mathematical and Computational Biology
>>>>>> +44(0)1582 763133 ext 2004
>>>>>> and...@bb...
>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>> --
>>>>> -ericP
>>>>
>>>> Andrea Splendiani
>>>> Senior Bioinformatics Scientist
>>>> Centre for Mathematical and Computational Biology
>>>> +44(0)1582 763133 ext 2004
>>>> and...@bb...
>>>>
>>>>
>>>>
>>>
>>> --
>>> -ericP
>>
>> Andrea Splendiani
>> Senior Bioinformatics Scientist
>> Centre for Mathematical and Computational Biology
>> +44(0)1582 763133 ext 2004
>> and...@bb...
>>
>>
>>
>>
>
> --
> -ericP
Andrea Splendiani
Senior Bioinformatics Scientist
Centre for Mathematical and Computational Biology
+44(0)1582 763133 ext 2004
and...@bb...
|