David Megginson writes:
>
>Curtis L. Olson writes:
>
> > This can significantly increase load times ... which is a hassle if
> > you are doing a lot of compile/run testing ...
>
>I just wrote a short ANSI C++ test program to read the airport
>database from the text file into an in-memory hash table. The basic
>data structures look like this:
I would like to make several observations about our 'static geographical
data'.
1) It is woefully incomplete
2) We are currently holding the whole world's data for some
data types in memory < Navaids in particullar >, this will
eventually become untenable as (1) changes
3) There are lot's of things we could add that would be useful
for example AM radio stations are a GOOD RDF source
which used to be used by a lot of navigators
4) Our current scenery bucketing scheme has several weaknesses
which make it unsuitable for general questions like
"give me all "airports within X distance"
ie it is based on an ~eaqual area but not a 'square area' tile
5) Several of our 'query functions" currently scan the entire DB
to find the "closest" point of interest. This will be a killer when the
databases get more fully populated.
6) FGFS is "global" in scope so we want to be able to access the
data for the whole earth from one database.
7) We only need to operate with a subset of the data at any time
for 'normal' purposes < things done every iteration >
ie data within some range
8) We want to access data outside of 'normal operating range'
occasionally ie goto airport
9) Some users might have 'personal data' that they want to use
and others might want to use the basic FGFS engine as a test
bed for various things with a geographical element
10) IMHO the same kind of 'generiticity' as is wanted for the rest of
the SIM as is evidenced by the popular 'rush' to property'ize
everything is just as germane for the 'external data' that we use
for Navaids, Airports ect.
Therefore given the above it makes sense to design a high enough level
interface to data that doesn't care whether the data lives in a diskbased
file or a 'in memory structure' in that eventually we will need a
combination
of both inorder to support the level of detail that we want !
Cheers
Norman
_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
