Click tolerence per layer

[ger-benjamin]

For GSGMF-1294

Variation from the specs

I've named the queryTouchMin option to toleranceTouch to be coherant with the tolerance for desktop and tolerancefor the edition (which is use basically for the the same behavior).
I've not seen any queryToleranceTouch to remove.

OpenLayers:

Done, PR here : openlayers/openlayers#11740
We have to wait on a new ol realease

C2cgeoportal:

It looks already done: https://github.com/camptocamp/c2cgeoportal/pull/6512/files
There is no tolerance or toleranceTouch in ngeoQueryOptions from constants of CONST_vars.yaml . And none is needed. A default value exists in ngeo (the same default value as before).

Ngeo:

I've take care of: #4570 (still and always send the simple click with tolerance bbox as GET parameter. A custom bbox can be done at the mapserver side, and I won't sent one bbox per layer).

[github-actions]

Examples: https://camptocamp.github.io/ngeo/click_tolerence_per_layer_gsgmf-1294/examples/
API documentation: https://camptocamp.github.io/ngeo/click_tolerence_per_layer_gsgmf-1294/apidoc/

[lgtm-com]

This pull request introduces 1 alert when merging 521c375 into 6c14ef1 - view on LGTM.com

new alerts:

• 1 for Useless conditional

[sbrunner]

Looks good but the CI should pass to make it possible to test it on pages :-)

[ger-benjamin]

So, we have to wait on an OpenLayers release.
If I target a commit in the package.json, I get some errors with the dist.

[ger-benjamin]

What's left is the ol update and this "invert position" requirement.
But I still don't understand this requirement.

First, what do you means by invert ? In the context, I understand flip vertically. it is correct ?
And by, The queryIconPosition specify the positions of the icon you means that the center of the icon is the center (centroid) of the queryIconPosition ? Right ?

Then, okay, the user knows the size of its icon. And he wants to apply the queryIconPosition from the center of this icon. So in this case, I've to shift (and not flip vertically) the queryIconPosition from the point (queryable point). And the amount of pixel to shift would be the computed bottom value of the queryIconPosition ?
It is correct ?

IMO That's complicated for the user. It's easier and more flexible to build the bounding box from the geometry point. But the specs are done, let me know what I've to do.

I don't see a scenario where to flip is the solution. (even if by invert you means flip horizontally, or reverse the bbox).

[arnaud-morvan]

Invert queryIconPosition : invert bottom and top and invert left and right.
That is, if the queryIconPosition represent the icon bbox relative to the geometry (geometry => icon).
To query the geometry after click in the icon we need to do the inverse (icon => geometry).

[ger-benjamin]

Thanks for the precisions. With this comment and a draw I've understand.
I'll invert the queryIconPosition values logic ( to [bottom, left, top, right] for 4, or,[bottom, left-right, top] for 3, or [bottom-top, left-right] for 2, or [all] for one value)

[arnaud-morvan]

Note that I'm not sure it is a good idea to mix bbox calculation from css definition, inversion and buffer application.
For comprehension, it might be better to do thing one by one.

[ger-benjamin]

@arnaud-morvan I've splitted the calculation but in the same function.
I hope it suits to you.

[arnaud-morvan]

Yes it suit to me, IMHO it is easier to understand now, thanks.