Currently we need to add the following (in viewer.js) to get redlining KML export working:

GeoExt.ux.LayerManagerExportWindow.prototype.baseUrl =
"${request.static_url('cartoriviera:static/lib/cgxp/sandbox/LayerManager/ux/')}";

Currently this is not documented at all.
We definitely need to find a workaround for this, or at least document it.

One solution would be to have a baseUrl config parameter for our redlining plugin, but this parameter need to be relayed to the LayerManager ux (which is n-levels deeper).

How to reproduce

• Select a theme (overlay) and activate alle it's layers
• Make sure no legend is displayed in the layertree
• Make sure the legend panel is close
• Monitor the http requests
• Zoom in the map
• The applications fires many unnecessary GetLegendgraphics requests

Expected

• GetLegendGraphics requests are fired only if the legend is activated in the layertree (for the specific layer) or if the legend panel is open

Get from #15

With IE I have noticed two problems with the layer opacity slider:

IE7 + IE8: the layer opacity is not always correct when playing with the opacity cursor: for instance the layer items seem darker on the map when slider value is 99% than 100%.

IE7 + IE8 + IE9: when playing with the slider cursor, the layer visibility sometimes switches to false (layer disappears) when sliding back to 100%.

Actually if we receive a feature without any attribute we add a new tab but without any columns ...

We should ignore feature without any attribute.

How to reproduce:

• Load http://map.cartoriviera.ch
• Click on 3d button
• Click on Drawing
• IE9 crashes

When the printing plugin is activated, the footprint should automatically be adapted (size+center) when zooming or panning. This behaviour is currently broken.

We need a live information on the azimuth angle and distance while drawing the circle, ie Lux : http://map.geoportail.lu/
Besides, as in LUX, we would like to have the altitude difference between the two points

The SITN project uses GeoExt.Lang.add to add translations. I don't think it should. In principle GeoExt.Lang.add should not be used in projects. Do we agree with that?

If we go at this URL the two layers (batiments_egid,batiments_projetes) are not in the correct order in the WMS query:
http://map.cartoriviera.ch/theme/Cadastre?map_x=555423.75&map_y=145277&map_zoom=6&tree_group_layers_Cadastre=batiments_egid%2Cbatiments_projetes&tree_groups=Cadastre

How to reproduce:

• log into the admin interface
• go to the layer section
• select a layer
• add the URL of a KML document in the field "KML 3D" and save
• load the webgis
• add the layer to the layertree
• a KML icon is displayed
• click on the KML icon
• nothing happens

Expected:

• click on the KML icon
• the KML is displayed on the map
• the GE plugin openitself and display the KML in 3D

1. request all layers even the ones not queryable:
   if a query is done on a group, all layers in that group are queried, even if the query has been deactivated in the mapfile, meaning a tab/bloc will be created for that layer in the result panel

2. in projet morges, the min/maxscale are not respected when doing a query: investigate to see if it's a general problem or only in morges

3. even if there is not result for a layer, selecting the related tab in the featureGrid trigger a hilight on the map: find why, there should be none.

Let's get rid of the accordion layout in the left region.

It would be nice to add some kind of actionConig property to the AddKMLFile and WMSBrowser plugins in order to define a custom tooltip for example (because I didn't find any way to define a tooltip on these plugins...). I think that Alex had done that for some other plugins...

On some machines, the connexion button doesn't work with IE. It does not depend on IE versions : on some machines it works with IE 8 and 9, and on other machines it does not work. To reproduce :

1°/ Open http://map.cartoriviera.ch in IE on a machine which has the problem
2°/ Connect with c2c/c2c -> The application reloads, but the connexion button is not in state connected

This is an important problem, several users have mentioned it, and I can reproduce it on my PC at home, but not on my PC at office, with the same IE 9 version...

THANKS

It would be great to have an API documentation that would list the CGXP plugins and detail their parameters. See for instance http://gxp.opengeo.org/master/doc/

when using the featureWindow plugin, hovering the mouse over the window's result hilight the related feature on the map.

but if this is done several time (ie, several query followed by several hover), 1) the hilight stop working and if we continue to try to move the mouse cursor over the window, 2) the window content disappear and become all grey.
at this point, the whole featureWindow plugin stop working correctly and it is not possible to do a new query.

this is visible on http://preprod.cartoriviera.ch

It would probably be better if the theme/layer search input was included in the theme selector. We could imagine highlighting the matching themes.
To be discussed.

in c2cgp when you enable draw (surlignage) feature, draw a point, a windows appears allowing the user to choose the color.

The label should be translated into the langage used in the interface. ie into french if french is a the default langage.

We just saw another bug in the edition plugin. If you don't activate any layer and click into the map, then the mouse pointer turns into the wait mode and doesn't come back to the normal pointer except if you click elsewhere...

If one define a layergroup containing layernodes that are all supposed to be initially unchecked, the layernodes and the layergroup are however rendered "checked".

It's OK if at least one of the layernodes is initially marked as "checked".

• Navigator : IE9
• clic on them selector
• choose a theme
• theme is added in c2cgp
• theme selector is not closed.

Feedback from a client :

I've added a new Layer Group. In there there are several Layers, some have a fill and some have a hatch. To draw the hatched ones over the filled ones I do have to give them an order greater than the filled ones. This way in the legend they appear below the filled layers but in the map they are drawn over the filled layers. This behaviour doesn't seem logically to me. The hatched layers should as well be above the filled layers in the legend (this would be the equal behaviour then as we have for the themes; the most upper ones are drawn over the other themes).

When printing through IE, the window telling that the print is done does not contain any text... I tried with SITN, geobl and cartoriviera and it always react the same.

Is that a bug or a Ext/CSS mis-configuration?

Either in the username field or in the password field, the "-", "_" and "+" keys are ignored. At least when the login panel is displayed above the map => in that case, pressing one of those keys will simply zoom/unzoom the map (no problem with the search field for instance). It seems that the key press event is intercepted by the map and not transmitted to the login field?

get from #15

A bug that is not new and occuring with all browsers: when moving up or down a layergroup, the background layer is switched back to the default one (or the first one in the combo?).

129bc0d#commitcomment-1696925

In CGXP we mainly use the OpenLayers.i18n() system whereas GXP uses GeoExt.Lang. Both systems are quite different:

• OpenLayers.i18n() is a simple key-value replacement system: strings are translated "on the fly" when used.
• GeoExt.Lang updates the i18n strings, directly in the properties of the GXP classes

Should we consider using GeoExt.Lang in CGXP as well, in order to keep the consistency between GXP and CGXP?

In addition this inconsistency may create problems. For instance with
https://github.com/camptocamp/cgxp/blob/master/core/src/script/CGXP/plugins/Measure.js
=> this plugin is inspired from a GXP plugin (using GeoExt.Lang) and does not use OpenLayers.i18n(). That would not be relevent since the interface strings are unchanged. However for now translations for this plugin are broken because GeoExt.Lang has no translation for it.
Should we change the plugin so that it extends the original one, and removing the interface string properties?

Hard coded here:
https://github.com/camptocamp/cgxp/blob/master/core/src/script/CGXP/plugins/FullTextSearch.js#L199

Asked by Cartoriviera

The code itself includes translation strings. For example, plugins/ContextualData.js includes this:

    defaultTpl: "Local Coordinates : {coord_x} {coord_y}<br />" +
        "WGS 84 : {wsg_x} {wsg_y}<br />",

This is overwritten in the en.js file:

    "cgxp.plugins.ContextualData.Tooltip.prototype": {
        defaultTpl: "Suisses Coord. : {coord_x} {coord_y}<br />" +
            "WGS 84 : {wsg_x} {wsg_y}<br />",

We should either use translation keys in the code, or use english in the code and remove translations in en.js. GeoExt uses english in the code, and I think we should the same.

If people agree I'll write a patch.

The checkboxes are not all checked in the displayed tree if the leaves of a group are set with "isChecked = true".

Currently a layergroup that has "isExpanded: True" remains collapsed in the layertree whereas it should be expanded.

When the Google Earth view is visible the user can use the keyboard to move the camera.

Unfortunately, the map also responds to these key events and so the map pans whenever the user moves the camera.

The map should not respond to keyboard events when it does not have the focus.

At https://github.com/camptocamp/cgxp/blob/master/core/src/script/CGXP/plugins/Print.js#L236

var printExport = this.target.tools[this.featureGridId].printExport();

=> The print plugin tries to access to the featureGrid plugin. If no featureGrid is available (for instance if one uses the FeaturesWindow plugin instead), a JS error is issued.

=> The print plugin should at least test if featureGrid is available before using it and if not, make sure that the PDF generation succeeds though. It would be useful to support other features listing tool (FeaturesWindow).

How to reproduce:

• Load http://dev.regiogis-beo.ch/sbrunner-thun/wsgi with IE9
• Click on Print
• try to rotate the print area on the map
• the map moves, there is no action on the print area overlay

(At least) in the bl_gis project, the WFSGetFeature control (and its action) automatically deactivates when a response is received from server.
If the response is empty or if no request is sent, the info tool stays active.

In my opinion, the same behavior (deactivation) is intended in both case.

If the feature grid already contains features and the user queries the map where there are no feature, the grid shoud be emptied.

Actually all the feature (of the FeatureGrid) will be send to the print, but most of them will be marked as display:none ...

Than those feature shouldn’t be send !

Some style definitions in https://github.com/camptocamp/cgxp/blob/master/core/src/theme/all.css shouldn't be defined in the cgxp css file.
div#header-out is a good example

Au sujet de #106 : on utilise plus souvent beforeDestroy plutôt que
d'enregistrer des listeners sur ses propres events. Voir
https://github.com/opengeo/gxp/blob/master/src/script/widgets/QueryPanel.js#L414
par exemple. Et on pourrait ajouter un petit test pour vérifier que
les listeners sont bien désenregistrés après destroy.

The vector layers of some CGXP plugins (Fulltextsearch, FeaturesWindow, RedLining ...) are sometimes stacked below WMS layers and then their features may be totally or partially hidden. It generally happens when the user add a new theme on the map.

In OpenLayers, I notice there is only a distinction between baselayers and overlays when setting the z-index of a newly added layer. Therefore, if one adds a WMS layer, its z-index will be higher than the ones of previously added vector layers.

I see at least one exception: it seems that the vector layer bound to the FeatureGrid plugin is always on top. I don't really know why (maybe because this vector layer is also linked to an OL.Control.SelectFeature or something like that, that would keep it above?).

We need to find a solution so that the application vector layers have always the highest z-indexes, no matter how many overlays are added meanwhile.

To pass styling options for instance.

Not really important... but there's a typo in https://github.com/camptocamp/cgxp/blob/master/core/src/script/CGXP/plugins/ContextualData.js#L109

I think "dans" should be removed...

when using the "X-UA-Compatible" content="IE=EmulateIE7; IE=EmulateIE9", checkbox behavior in layertree is broken in IE8

checking/unchecking may require several clic to be working instead of only one

We don't have continuous testing in place for CGXP. And It's unlikely to happen anytime soon. In the mean time we should probably put the test runner page online somwhere, so we can run tests at any time without the need for a local installation.

He should use the gxp method:

https://github.com/opengeo/gxp/blob/master/src/script/plugins/Tool.js#L309
https://github.com/opengeo/gxp/blob/master/src/script/plugins/Tool.js#L354

CU
Stéph

When doing consecutive queries on the map (using the query tool) the following javascript error is raised:

TypeError: store is null Featur...odel.js (line 254)

Looking at the trace, it seems to be related to an error on line 557 in FeatureGrid.js

[ContextualData]
Currently we're using showLocationInMapRequestonGoing to prevent new requests to be sent if a there's a request for which we're waiting for a response.
I think that the resulting behavior is disturbing for the end user when using mouseover. In this case, it's almost impossible to know for which position the response from server is given.
Rather, I think it would be better to cancel ongoing requests if a new request is to be sent.

If not specified, projectionCodes should be the map projection.