ld_dir(...) using graph URI in global.graph even when specified explicitly

[randykerber]

I'm passing a ld_dir( dir, filePattern, graphUri ) command into isql.

In the directory I am loading data from, there is a global.graph file with the default graph
URI ("http://rdf.ncats.nih.gov/opddr") for data files in that directory.

Even though in the ld_dir() commands passed in I am specifying the third argument, Virtuoso is
ignoring that argument and instead using the value in the 'global.graph' file.

This happens even for the file named '2015 pubchem_pd2_assay.ttl.gz' where there is a '2015
pubchem_pd2_assay.ttl.gz.graph' file with the correct graph URI.

The same behavior is seen both in Ubuntu and with virtuoso running in docker on MacOS.

Below is some sample output from a bash shell:

Virtuoso Open Source Edition (Column Store) (multi threaded)
Version 7.2.4.2.3217-pthreads as of Mar 17 2017
Compiled for Linux (x86_64-unknown-linux-gnu)
Copyright (C) 1998-2016 OpenLink Software

---

ubuntu$ ls -l /staging/ncats/data

-rw-rw-r-- 1 ubuntu ubuntu    30 Jan 14  2016 global.graph
-rw-rw-r-- 1 ubuntu ubuntu   848 Oct 13  2015 npcpd2_assay.ttl.gz
-rw-rw-r-- 1 ubuntu ubuntu  1034 Oct 13  2015 npcpd2_bao.ttl.gz
-rw-rw-r-- 1 ubuntu ubuntu 21460 Oct 13  2015 npcpd2_substance.ttl.gz
-rw-rw-r-- 1 ubuntu ubuntu  1442 Oct 13  2015 pubchem_pd2_assay.ttl.gz
-rw-rw-r-- 1 ubuntu ubuntu    38 Jan 14  2016 pubchem_pd2_assay.ttl.gz.graph

ubuntu$ cat /staging/ncats/data/global.graph
http://rdf.ncats.nih.gov/opddr

ubuntu$ cat /staging/ncats/data/pubchem_pd2_assay.ttl.gz.graph
http://rdf.ncats.nih.gov/opddr/pubchem

before calling 'ld_dir()', no entries in DB.DBA.load_list table for files in /staging/ncats/data.

ubuntu$ isql 1111 dba dba exec="select ll_file, ll_graph from DB.DBA.load_list where ll_file like '/staging/ncats%';"

Connected to OpenLink Virtuoso
Driver: 07.20.3217 OpenLink Virtuoso ODBC Driver
OpenLink Interactive SQL (Virtuoso), version 0.9849b.
Type HELP; for help and EXIT; to exit.
ll_file                                                                           ll_graph
VARCHAR NOT NULL                                                                  VARCHAR
_______________________________________________________________________________


0 Rows. -- 0 msec.

ubuntu$ isql 1111 dba dba exec="ld_dir( '/staging/ncats/data', 'pubchem_pd2_assay.ttl.gz', 'http://rdf.ncats.nih.gov/opddr/pubchem' );"

Done. -- 3 msec.

ubuntu$ isql 1111 dba dba exec="ld_dir( '/staging/ncats/data', 'npcpd2_substance.ttl.gz', 'http://rdf.ncats.nih.gov/opddr/substance' );"

Done. -- 0 msec.

Both files loaded into URI in the global.graph file despite URIs passed to 'ld_dir()'

ubuntu$ isql 1111 dba dba exec="select ll_file, ll_graph from DB.DBA.load_list where ll_file like '/staging/ncats%';"

ll_file                                               ll_graph
_______________________________________________________________________________

/staging/ncats/data/npcpd2_substance.ttl.gz           http://rdf.ncats.nih.gov/opddr
/staging/ncats/data/pubchem_pd2_assay.ttl.gz          http://rdf.ncats.nih.gov/opddr

2 Rows. -- 1 msec.

[HughWilliams]

This is expected behaviour as detailed in the docs at:

https://virtuoso.openlinksw.com/dataspace/doc/dav/wiki/Main/VirtBulkRDFLoader#Bulk loading process

ie .graph file take precedences over the graph IRI specified in ld_dir() ...

Note also the pubchem_pd2_assay.ttl.gz.graph file name you specify is incorrect and should be pubchem_pd2_assay.ttl.graph i.e. ignore the .gz compression extension as it will be uncompress for loading at which point its name will be pubchem_pd2_assay.ttl ... and is why its contents were not used when ld_dir() was run in the example you gave ...