[OpenLayers-Dev] Layer.Vector: getFeatureBy and getFeaturesByAttribute

[Marc Jansen]

Hi Eric,

thanks for your comments. I've added my thoughts inline:


On 11.12.2010 22:29, Eric Lemoine wrote:
> On Thu, Dec 9, 2010 at 6:16 PM, Marc Jansen<jansen at terrestris.de>  wrote:
>> Hi list,
> Hi Marc
>
>> instances of OpenLayers.Vector have a method getFeatureBy(property, value)
>> that was introduced in revision 10691 [1] and lives in 2.10. This method
>> returns the first feature that has the property set to the given value.
>>
>> Wouldn't it make more sense to have a function getFeaturesBy(property,
>> value) that returns an array of matching features?
> I think getFeatureBy was just introduced for the purpose of
> getFeatureById and getFeatureByFid, and id and fid are primary keys
> here. I'm even wondering if it should be part of the API.

Would you prefer hidding those methods from the API and let all 
application code work with the "features"-attribute of Vector-layers? 
The "feature"-array isn't an API-property either, is it?

>> AFAICT vector layers also do not provide a method to get features that have
>> a certain attribute set to a given value. I propose a new method
>> getFeaturesByAttribute(attrName, attrValue, strict) that would scan the
>> attributes of features for matching candidates and eventually return those
>> as an array (see [2] for a hardly tested implementation).
>>
>> What do you think?
> I would personally not add such a function to the library. My reason:
> coding this function is easy enough, so I'd leave its implementation
> to the application developer, thereby avoiding bloating OpenLayers
> with code that most application developers won't need. Also, by having
> this function in the application code, you will probably not need the
> "strict" argument and the logic that goes with it in the core of the
> function, so this will result in fewer bytes of JavaScript, overall.
>

I have found myself implementing different versions of the function 
myself... which is why I thought it would be better suited as reviewed 
method inside the library and not in the application code. Furthermore I 
think that the library should provide a method of getting features by 
their attributes, and that application-code has to be implemented to 
work on that result set of features. These are just my thoughts -- I can 
fully understand that others might have different thoughts on this topic.


> Just my personal view of things.
>

Thanks again for your thoughts.

Kind regards,

Marc