Martin Desruisseaux ha scritto:
> I'm sorry that some topics has gone that emotional. This is true that 
> figthing 
> against complexity is a very hard topic - hacking GeoTools is a daily fight 
> against entropy.
> 
> On my side, the complexity in GeoTools that scare me the most is not the ISO 
> model complexity. While ISO models try to scope with a complex reality, they 
> usually do that in a coherent way (in my feeling). I means their mathematic 
> are 
> coherent (as opposed to "intuitive" trick that the computer field was used to 
> do, e.g. in date handling, but which were mathematical discontuinity).
> 
> What scare me in GeoTools is rather the lack of cross-module consistency. 
> Same 
> code duplicated, calculation for the same things performed in different way 
> (AffineTransform in some modules, explicit scale/translation in other 
> modules) 
> which usually leads to different results, etc. I see them as complexity which 
> scare me more than ISO models complexity, because ISO is consistent while 
> lack 
> of cross-module consistency is a cause of hard-to-predict behavior...

I agree that lack of consistency is a liability. But at the same time, I 
would like you to take 5 minutes and skim over the WFS specification, 
just to have a feel of size and complexity of it: you'll notice that the
spec is not that big and not particularly complex (thought it builds
on the gml one, which is quite a bit bigger).

Why did I cite WFS you ask? Well, because on gs-users we daily manage 
questions whose answer is in the WFS spec itself, mostly because:
* people did not read it
* tried, but were put away by the dry presentation of it
* tried, but did not understand it

The point I'm trying to make is that buying 100% of big specs like the
ISO geometry model may make GeoTools usable only by the very few people
that managed to understand the spec.

Consistency is not a direct consequence of spec adherence in my experience.
I've been talking with people trying to make WFS clients for servers
that do implement the WFS spec and pass the CITE tests. Guess what,
despite that, make a client that follows only the spec, and you won't
be able to speak with any of the existing servers out of the box,
some tweaking (different for each server) is needed to make the
communication actually succeed. In fact, if you look in our WFS module,
there are server specific "strategy" objects that are there to handle
the specific issues of communicating with each server.

Lack of consistency in GeoTools is, in my opinion, not the
result of not following a spec, but the result of having multiple
people work on different modules, in different times, without
a central master mind to coordinate and decide each move.

Cheers
Andrea


-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft 
Defy all challenges. Microsoft(R) Visual Studio 2008. 
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Geotools-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/geotools-devel

Reply via email to