gem / oq-platform Goto Github PK

We would like to implement the following features:

• if you click on one region in the map, toggle its highlighting in the charts and in the tables (highlight it if it was not highlighted, or unhighlight it if it was highlighted)
• if you highlight one or more regions in a chart
  • highlight it/them in the other charts
  • highlight it/them in the map
• consecutive clicks on different regions add the clicked regions to the selection (and to the highlighting)
• SHIFT-click draws a bounding box and selects all features inside (highlighting them in the map)

The features implemented in #594 (addressing #596) require additional selenium tests.
In particular, we want to test that:

1. when a zone is clicked in the map, the corresponding line is toggled (highlighted or unhighlighted) in all charts. Consecutive clicks on different zones make all those zones highlighted in the charts.

2. when brushes are used to filter data in one chart, the selected zones are also highlighted in the other charts.

• OQ logo misplaced
• menu items bottom offset wrong
• username sub-menu arrow is too low
• the main content is slightly shifted to the left (see the tile in the screenshots)
• the footer is broken (too narrow)

should be:

The result is the same with FF and Chrome

Reported by @pslh, a link to this page http://www.globalquakemodel.org/what/physical-integrated-risk/building-taxonomy/ should be present in the taxtweb app

The style is still based on the previous OQP design. This app should moved to the current design. I would in any case do this after the migration to GeoNode 2.4 which uses bootstrap by default for the core.

limit_states: any not letter or number must be converted to '', multiple '' are collapsed.

mix of continuous and discrete fragility functions as NRML file is denied

We are currently checking all the three version numbers, major, minor and maintenance: version 1.7.1 of the plugin is not working because we are looking for COMPATIBILITY_VERSION = 1.7.0.

We should check up to the second number (minor) only, since in any case no changes in the protocol should be expected without increasing the minor number.

COMPATIBILITY_VERSION = 1.7.0
1.6.x -> KO
1.7.0 -> OK
1.7.1 -> OK
...
1.7.x -> OK

We should also decide what to do if the minor number is greater, like 1.8.0 is this fine or not if the COMPATIBILITY_VERSION is 1.7.0 ? (I vote NO for the current status, but it should be YES)

When generating an exposure model using the aggregated value option, and extra “ is written in the output file. I used the attached exposure information (exposure_example.txt) which led to the exposure_example.xml also attached.

In RIPT copy and paste does not work if any dropdown menu is changed and a column is added. It work when the page loads instead.

How to reproduce:
1 Load the RIPT app
2 Change the first dropdown available (Structural Costs) to any choice (like 'Aggregated')
3 Copy sample data from Excel o LibreOffice (like a b c d e f)
4 Try to paste those values into the table. Nothing happens

It works is step 2 is skipped.
This is reproducible on the live Platform (v1.3.1 as writing) and has been reported by Desmond.

Currently the IRI values are being computed using scaled SVI and scaled RI values. We need to compute IRI using SVI and RI native values. And continue to pass to the map and to the charts the scaled values.

We need to manage the Bing API key during the installation:

1. if a Bing key is not provided by the user we need to delete the Bing basemap layer from the pre-loaded fixtures
2. if a Bing key is provided we need to update the fixture with the valid key

About the local_settings.py we should manage it with an if

Remove Bing from MAP_BASELAYERS 
then if BING_KEY is defined
    append the Bing basemap with the BING_KEY to MAP_BASELAYERS

See also #484

To test it remove LoginRequiredMiddleware item in MIDDLEWARE_CLASSES tuple in local_settings.py.

MIDDLEWARE_CLASSES = (
    ....
    # geonode.security is enabled to lockdown the website
    'geonode.security.middleware.LoginRequiredMiddleware',
)

Traceback:
File "/usr/lib/python2.7/dist-packages/django/core/handlers/base.py" in get_response
  115.                         response = callback(request, *callback_args, **callback_kwargs)
File "/usr/local/lib/python2.7/dist-packages/openquakeplatform/vulnerability/views.py" in list
  72.     user = User.objects.get(username=request.user.username)
File "/usr/lib/python2.7/dist-packages/django/db/models/manager.py" in get
  143.         return self.get_query_set().get(*args, **kwargs)
File "/usr/lib/python2.7/dist-packages/django/db/models/query.py" in get
  404.                 self.model._meta.object_name)

See:

• http://devblog.mapquest.com/2016/06/15/modernization-of-mapquest-results-in-changes-to-open-tile-access/
• http://lists.osgeo.org/pipermail/geonode-devel/2016-July/001064.html and following

It would be important to unify all calculations within one single function to be called recursively node-by-node on the project definition tree.
It would: 1) avoid duplication; 2) make the tree structure flexible (currently calculations assume a fixed tree structure).

We recently switched the Python server on production from uWSGI to Gunicorn due issues that we experienced with uWSGI and POST in the Vulnerability app (empty download cart). We'll need to update the documentation about uWSGI/Gunicorn.

We need also to do more testing using Gunicorn (by the way Icebox, which is the most critical piece, looks working well).

In the Risk Preparation Tool after playing with the drop down menus, when you try to add a new row (with ctrl-shift-n) more than one row is added (equal to the number of fields changed). It looks like more a bug in the library: the .destroy() method (fired when a menu option is changed: https://github.com/gem/oq-platform/blob/risk-ipt-bug3/openquakeplatform/openquakeplatform/static/js/risk_input_model_exposure.js#L121) is not completely removing the table from the DOM, so the .keyup listener iterates on N tables (https://github.com/gem/oq-platform/blob/risk-ipt-bug3/openquakeplatform/openquakeplatform/static/js/risk_input_model_exposure.js#L204).

Elements of group of material technologies (mat_tech_grp) are wrongly assigned.

in the exposure.xml the opening < bracket is missing.

Add a check in the tool for:
"It is not valid to define a loss ratio = 0.0 with a corresponding coeff. of variation > 0.0" @bwyss

Although it looks like the data is re-processed, the map does not display any change.

Currently, when no project is selected, no information is displayed when clicking on the map (which is fine). When a project is selected and you click on the map, a box is displayed showing information about the clicked feature. The problem is that we should display only information that is related to the project, ignoring everything else. Currently, also irrelevant information is shown, such as names of the region provided in different languages, etc.

Currently by default the currency is “EUR”

When running the taxtweb app on the stand alone URL (http://taxtweb.openquake.org) the menu located in the header does not play well:

• To make the calculate / share / explore buttons working a change in the Nginx (or other reverse proxy) conf must be done. We need to document it.

The issue is caused by:

        location / {
                rewrite / /taxtweb/ break;

and can be fixed with a new location block

    location ~* /(calculate|explore|share|account\/signup|account\/login)/$ {
        rewrite        ^ https://platform.openquake.org$request_uri? permanent;
    }

• The sign in button (or the missing username + avatar) should be removed or hide when in the stand-alone URL since the authentication cookie is not recognized across different domains.
  The register button could be left there, but the fix above must be applied also for it.

The only way to remove supplemental layers, once they have been added to the map, is to reload the page.
This should be probably addressed together with #525

Currently, if the input exposure data does not include the number of occupants, i.e., if the Day, Night, and Transit checkboxes for Occupants in the input form are all left unchecked, the generated NRML still includes an empty occupancies tag for each asset:

<asset id="EUCENTRE-3" number="1" taxonomy="Concrete" >
    <location lon="9.1582" lat="45.18472"></location>
    <costs>
        <cost type="structural" value="6000000"></cost>
    </costs>
    <occupancies>
    </occupancies>
</asset>

This should be avoided.

Chrome says:

Should be

When a layer is selected from the "Supplemental Layers" dropdown menu, it is added to the map. If the user then switches to a different layer, also that second one is added to the map, together with the first one. If afterwards the first one is re-selected, it is re-added to the map, instead of just re-selecting the existing one.
The fact that a simple dropdown selection enables to add multiple options to the map looks a little counter-intuitive. If we really want to enable multiple selections, we should probably do so by means of a mutliselect widget.

The error is

WARNING: curl -v -s -o output/styles/qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld -D output/styles/qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld.hea -u admin:geoserver -XGET http://127.0.0.1:8080/geoserver/rest/styles/qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld
WARNING: web_cmd curl GET qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld  http://127.0.0.1:8080/geoserver/rest/styles/qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld 200
         expect 200, receive 500 (log into: output/styles/qgis_svir_30eadaa4_245a_4eca_911a_6731ad8d27c5.sld[.hea])

We are investigating the issue in GeoServer.

When a project is loaded in the IRV, the titles of the single project definitions belonging to the project are visible inside a dropdown menu (used to select a project definition from the list).
It would be good to display also the main title of the project somewhere. An easy solution (perhaps sub-optimal) would be to place it on top of the project definition widget.

If geographical zones are displayed as x-labels and geometries are available, we could sort zones based on geographic proximity (calculating distance between centroids), attempting to keep close to each other in the chart the zones that are geographically close (instead of ordering alphabetically).

For instance, in the composite indicator chart the user may want to highlight what are the zones that score high on a particular indicator. In the current implementation, we can do the opposite: highlighting indicators that score high in a particular zone.
The composite indicator widget could have 2 tabs, containing the two options. The same can be done for the other widgets.

Using the Exposure export to download data from Greece and Portugal and maybe more gives an error.

How to reproduce:

• Load the Exposure export and select Load data by region
• Filter to 'Greece'
• Select 'Greece | PAGER'
• Try to export 'Gridded Building Exposure' + 'Residential' + 'CSV'

You'll get a new web page wit this error:

The export can not be performed because the total grid count for study region id xyz exceeds the threshold (NNNNN > 100000)

You should get the possibility of drawing a box on the map to reduce the exported area instead (like Italy).

Thanks to @monellid for discovering the bug: https://groups.google.com/forum/#!topic/openquake-users/a4rlxtAYFIs

Add to the current charts the possibility to do like in http://bl.ocks.org/mbostock/1341021
This would make possible, for instance, to highlight zones that score within a specific range for one or more indicators. Or, if axes are inverted, it could enable to highlight indicators that score in a given range for one or more countries.

Currently, if there is some missing data in the layer, the IRV application doesn't show anything in the map, and neither in the charts. It only shows the following message: The application is not able to render charts or the map layer for this project because the composite indicator data is incomplete.. Furthermore, such message appears when the project is loaded, but it disappears afterwards, for instance when a widget is hidden and shown again.
We definitely need to handle missing data properly, showing meaningful results for all the features for which primary and composite indicators can be calculated.

Currently by default the description is <description>exposure model</description>

I think is correct to have a default value, but users should be able to modify description if need it

The IRV leaflet thematic mapping function expects IRI, RI, and SVI values in every project. Thus the default thematic mapping menu option is IRI, and when there is not IRI value in the project, the map is rendered incorrectly.

The IRV web application does not manage negative theme indicator values correctly.
It seems that the sign of the original data is being switched at some point of computation.

When the map is styled with respect to an indicator (or composite indicator) that has a missing value for one of the regions, such region is (correctly) not colored. By clicking such region, currently the application displays no data available, which is true with respect to the selected indicator, but a bit misleading in general. It would be better to display anyway the whole set of data available for that region, regardless from styling.

The fragility and vulnerability functions on the platform are still exported in the old NRML format and should be updated to the new one (NRML 0.5).

1. In some cases, the check on the existence of the new project definition name fails, saying it already existed although it did not. It does not happen always, and probably it's due to the fact that some events can occur with different delays, causing different behavior depending on the order things happen.
2. When a new project definition is added, it is also pre-selected from the dropdown menu. If afterwards another one is selected, and then the new one is re-selected, it can not be loaded. The only way to be able to use it afterwards is reloading the whole page, which makes the new item actually available.
3. I've also noticed a redundant check on the existence of an assigned name for a new project definition. The check is already done by the API that is called by the IRV app, so there is no need to repeat the same check here.
4. In some cases, the addition of a new project definition is performed multiple times redundantly, instead of triggering it just when the submit button is clicked. This seems to be due to unnecessarily nested event definitions.
5. The code contains some unnecessary delays, that were added to partially fix some of the above issues, but that should be removed once the real causes of these issues are fixed.

We need to put the Platform feature history in the CHANGELOG, at least from v1.0.0

Even if you use mouse wheel or "+"/"-" buttons zoom stops to update the map and the zoom cursor.

The webGlColorPicker menu seems to work at first, but on my computer (both using firefox and chrome) if I switch to different colors many times, the response time approximately doubles at each attempt. It is very quick at first but, after about 6 consecutive changes, the response becomes quite slow and it causes my firefox to open the following warning popup:

A script on this page may be busy, or it may have stopped responding. You can stop the script now, open the script in the debugger, or let the script continue.

Script: https://api.tiles.mapbox.com/m…box-gl-js/v0.11.1/mapbox-gl.js:194

I guess we should also decide if the possibility to change colors is actually a useful feature.

• when too many indicators are shown, by default the widget is too small to contain the unwrapped table
• the line for null values might be thicker and the null values might go lower
• the (displaying n of m) string would probably look better below the table
• the number of items displayed in the table should depend on the amount of space available in the widget below the chart
• resizing the widget, the table should not overlap the chart

When a zone is clicked on the map, we want to display information in a more readable and structured way, dividing variables by section (e.g. keeping in a same section all the primary indicators) and displaying bar charts instead of numbers.
See http://www.cs.umd.edu/~bederson/papers/index.html as a possible reference.