Cannot browse larger data

[jakubklimek]

Repro: Vykaz zisku a ztraty, T-SPARQL 2014, output, browse

[jakubklimek]

fixed in current develop

[jakubklimek]

Happening again with [MFCR] Rozvaha 2010-2014 Year by Year, T-SPARQL 1 2010, output, browse

[skrchnavy]

Outstanding issues:

1. optimising (construct) sparql query for size information in: RDFQuery.getSizeForConstruct(RDFDataUnit rdfDataUnit, String constructQuery) concretely line 335-336 Model model = QueryResults.asModel(result); size = model.size(); initialises the whole model just to get the size of it and then throws the result away with big data this slows debugging tremendously
2. this is called for example by RDFQueryView.runQuery() : line 364 calls: if (container.size() > 0) ... just to get the info if its empty. Can it be replaced with optimised isEmpty method?

refer to commit 3e377cb710a3ccf87e1f69caca22330e4443668c

[ghost]

1. resolved by mvi, iterating graph result. not optimal, nothing better can be done, maybe

  final String sizeQuery = String.format(
                "%s SELECT (count(*) AS ?%s) WHERE {%s}", queryPart
                .getQueryPrefixes(), sizeVar,
                queryPart.getQueryWithoutPrefixes());

this is done for select, mangling query to obtain size directly from database. Someone who is SPARQL-pro can tell if something like this can be done for CONSTRUCT too.

2. is the size of contruct even needed? why? let the user see empty table. solved. this size-obsession is crazy. lets get rid of these size checks from frontend, just show the output to the user (or empty table when there are no triples satisfying query) and do not ask for size at all!!

[tomas-knap]

I guess that size is used to know how many pages with the data Vaadin should depict.

[skrchnavy]

This issue is closed, rest will be fixed as #220