Geometry Boost List: Re: [geometry] Some wild rtree ideas

Subject: Re: [geometry] Some wild rtree ideas
From: Adam Wulkiewicz (adam.wulkiewicz_at_[hidden])
Date: 2014-11-25 19:45:07

Hi Henry,

Henry Roeland wrote:
> Dear all,
>
> I have put a question on stackoverflow 
> http://stackoverflow.com/questions/27040161/boostgeometry-rtree-and-paging-lazy-load/27072352#27072352
> Concerning boost geometry rtree and loading pages when needed. The 
> answer is that iT is not supported yet.
> http://stackoverflow.com/users/2574249/adam-wulkiewicz pointed me to 
> this mailing list for help and ideas.
> So hier I am.
>

Sorry that I didn't answer earlier. It's great that you joined us :)

> Why page loading?
> Some maps are to big to index as a whole and would-be beter to load 
> parts only when needed.
>

Of course, as mentioned in the other thread (Crash in rtree after 
growing shared memory, VS2012 x64, Boost 1.55) the rtree should be able 
to store data in a persistent storage one way or another. One of the 
main purposes of this data structure is to allow storing data in a 
persistent way.

> Other ideas:
>
> In combination with boost interproc maybe support for compressed 
> mapped files? lz4 is quick enough when reading.
>
> Custom page load mechanism that supports:
> - retrieving data from different data sources ( database, zeromq,.... )
> - dependcy checker to unload a page ( file, db query, time,....)
>
> I realy have no idea if above suggestions are possible.

Everything is possible :)

Yes, a designed solution should be as generic/universal as possible and 
allow usage of various storage options.

I have some ideas which I'll send tomorrow since I don't have free time 
right now. I also saw your other email, which seems to be a better place 
for more detailed answer. I'm sure we'll figure something out.

Regards,
Adam