[R-sig-Geo] Qgis-Rmanage and geographic display of spatial objects

[Carson Farmer]

Agustin Lobo wrote:
> Carson Farmer wrote:
>> This is certainly where I would like to see manageR go someday! But 
>> unfortunately this would entail programming in C, which requires 
>> someone much more skilled than myself!
>> For now I simply use it instead of jumping between the two 
>> applications...
>> Your suggestion about displaying sp objects was a topic of discussion 
>> for the QGIS bridge2R team (of which I am a member), but we ended up 
>> putting this aside, as we felt that the average QGIS user would 
>> rather have all the R stuff happen 'behind the scenes' on their input 
>> vector layers directly, rather than knowingly interacting with R.  
>
> I cannot but disagree here. "QGIS user have all the R stuff happen 
> 'behind the scenes'" because they cannot have it on the scenario. I 
> strongly
> recommend having a look to GeoDa to feel the power of interactive
> exploratory geographic data analysis. It is a fundamental step for 
> thinking and generating hypothesis that are formally tested afterwards
> and then interactively displayed again. Unfortunately, GeoDa does not
> have a real geographic display, as would be the case of QGis, and does
> not have the amazing flexibility of R. I note
> that GeoDa is becoming OS at some point,  and could be an excellent 
> source of inspiration.
Reading over my last post, I can see that I wasn't clear on this point.
What I meant to say was: Brushing and Linking is good, and would be very 
nice to have (and in fact I do have a graphing plugin that does this in 
the works as I write), but we are talking about two separate things here:
1) The bridge2R project; the primary goal here (I think, other people 
can correct me here) is to provide a user-friendly means of doing 
statistical analysis in QGIS. The fact that we are using R is simply 
secondary (though the flexibility and extensibility of R means that we 
can continue to add functionality to QGIS long into the future), and 
thus this project should provide statistical operations to non-R-users 
and R-users alike. This means that brushing and linking can be 
incorporated into these operations, but the user shouldn't know that 
it's R and Rpy doing all the work.
2) The second point is something like manageR, or some other direct link 
between R and QGIS.  I imagine this as something of interest to more 
advanced R-users, who want to be able to knowingly work in R, and 
interact with QGIS at the same time. This is what I am most interested 
in, and am looking for ways to make manageR closer to this type of tool. 
This could also support brushing and linking eventually...?
>
> Also, note how powerful would be applying R syntaxis to the data 
> tables associated to vector data. This is very easy with the "data 
> slot" of R
> SpatialDataFrame objects, but there is always the annoying step of going
> through the shp export to display in QGIS (although I think that your
> Rmanage makes easier the step).
and hopefully I made this even easier with my latest update.

> Actually, perhaps we could make people from Rosuda get interested on
> developing an "igeoplots" package!
Their website indicates that there will be interactive mapping abilities 
(based on maptools) soon-ish...

Carson