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... |