cpvrlab / sparql-table-viewer Goto Github PK

The scroll bar has the wrong dimensions after running the query for the first time.

See how to optimize performance with preloading of the first queries as also with the generated queries themselfs.

https://github.com/zazukoians/sparql-http-client

Right now while scrolling just at the moment the user hits the not loaded space a new page is loaded. It should be possible to anticipated that and preload a page.

This will not work for sudden jumps obviously.

It's currently not possible (from what I saw) to share a certain filter with someone else. It would be nice if I can get a persistent URI for a certain configuration and share this one with other people.

If a filter is active (not everything selected) then indicate that to the user by colorizing the icon.

There is this problem that there are columns in the input file which are bound to one specific column which is a key of the dataset. In UBD66 this is the SitePlot one. Currently this is simply shown as the unique key for this column.

After talking to BAFU I had to add the remaining columns in a way that we can query them as well. For that I did two things:

• SitePlot is now a skos:Concept and it points to additional "pseudo-dimensions" via skos:narrower. Each of this dimensions is again a skos:Concept and contains a skos:prefLabel.
• Each pseudo-dimension is also its own skos:Concept and to this definition I attached a qb:order so we can figure out which one should be shown first.

Example:

• http://environment.data.admin.ch/ubd/66/siteplot/31_1 - points to all the skos:narrower
• http://environment.data.admin.ch/ubd/66/altitude/770 - points to the class Altitude
• http://environment.data.admin.ch/ubd/66/qb/Altitude - class which contains the qb:order

I hope the idea is clear, if not please let me know.

http://test.lindas-data.ch/sparql?default-graph-uri=&query=select+distinct+*+where+%7B%3Fa+%3Fb+%3Fc+.+%7D+LIMIT+1000&should-sponge=&format=application%2Fsparql-results%2Bxml&timeout=0

The current filter drop down shows a checkbox for every distinct value found in that column. This is ok for literal types like names, but it introduces a few problems with number types.
namely it leads to very large sparql queries since every checkbox adds a filter to the query. It would be great to have a range filter for number types.

Reimplement the download functionality with content-negotation.

First, the viewer is really great IMHO, very easy to create customized tables, thanks for the work!

What confused me was that once I did start to filter I couldn't easily reset all filters so I think a "clear filter" button or anything similar would be useful.

To get all values of one row just wrap the SPARQL query in another one like below. This gives you all possible values to filter for.

SELECT DISTINCT(?Field_name) WHERE {
    SELECT * WHERE { ... }
}

Sometimes 0.0 values in the measurement column get displayed as NAN. This seems to be an issue with slickgrid.

Reproduce: keep sorting the measurement column until you see NAN values.

With lindas-uc/bafu_ubd@975f475 I've added qb:order to the Data Cube definition.

Please note that this is optional so the viewer should also work without.

http://answers.semanticweb.com/questions/20635/return-only-the-first-match-in-a-sparql-language-filter

If a filter is already active reload the other filter contents to only include options that are still in the current list.

Possible problems:
If a selection on a filter is already active and we change a second filter, we must reload the content of the former without losing its selection.

The user is able to click on a header cell to sort that column, after having just clicked on the same header cell and while the data for that previous click is being updated. The sort arrow on the header cell will still switch but the data of the previous click will be loaded, leaving the arrow pointing in the wrong direction.

It's just a visual bug, by clicking again the correct sort direction will be used.

I now implemented Data Cube vocabulary for the BAFU UBD28 dataset, you can find more information in the corresponding issue.

If I'm not mistaken it should contain all information for creating the view dynamically, in terms of all stings provided on the demo page should be contained in the metadata.

Two things happen if we change filter checkboxes while data is being loaded:

1. the loading indicator vanishes.
2. the new new filter we just set might not be applied until a new data load is triggered.

Should be solved by adding the LIMIT and OFFSET by using a SELECT wrapping.

SELECT *
WHERE
  {
    {
      SELECT * WHERE { ... }
      ORDER BY ASC(?p)
    }
  }
OFFSET 50000
LIMIT 1000

Add a title field for Pivot Tables so that data fields that are uniform across the table can be dragged into the title field.

Example: http://yasgui.org/short/4J3fgHMhx
In this example, pollutantValue, aggregationValue, and unitValue should go into the title field and be displayed on top of the table.