Re: [Geotools-devel] 3 geometry bugs I'm looking into

[Graham D. <gd...@re...>]

Jackson,

Glad to hear from you, and it will be nice to get some of your help with 
these bugs.  I think Jody answered your 2 questions pretty well already.

If you are planning to look at some of these bugs on the weekend, I'll 
be sure to email you on Friday before I leave work to give you an update 
on where I am at with them.  I am currently looking at the relate() bug 
and have been deep in the debugging of it, comparing it with the 
original JTS code since it is based on that.  So it is probably best if 
you look at one of the other 2 bugs since I am already deep into this 
one.  However, if you have any insight or ideas about this particular 
bug, please email me anything that comes to mind (so far this debugging 
has got me in loops!).

I look forward to collaborating with you on this module,

Graham.


Jody Garnett wrote:

> Prof. Dr.-Ing. J. Roehrig wrote:
>
>> Dear Graham, dear Jody
>>
>> As Jody said yesterday, the implementation is not yet working.
>>
>> I will have some time to work on it again (until September) and try 
>> to solve some bugs. But before I go ahead, I would like to ask you 
>> two important questions for me:
>>
>> 1) In your opinion, what are the chances that the implementation will 
>> be adopted (used) in the future, if we solve this serious bugs? If 
>> another implementation will be adopted (e.g. a wrapper), I would 
>> prefer to invest my time in other topics. If the current 
>> implementation is intended to be improved and adopted in the future, 
>> I think I can find a student to help us.
>
> The chances are 100%:
> - we have done the initial pass of testing showing that only three 
> major bugs exist
> - we are being paid to make this implementation adopted and used
>
> Specifically we need to bring your module up to supported status (a 
> formal part of the GeoTools library). It is already a success as far 
> as a consulting company is concerned. I have talked with Martin Davis 
> about ways in which we could combine forces with JTS in the future - 
> providing we can find funding for such an effort.
>
> Aside: Discussion has focused on isolating SFSQL interfaces for 
> GeoAPI, and focusing on making a "lookup or stratagy" based system for 
> the topological operations. Martin is only interested in the 
> primitives with straight lines, we could recognize that case and make 
> use of the JTS implementation as an optimization in this special case. 
> For working with Curves or 3D we would be making use of code in your 
> module - and I agree there is plenty of work to do.
>
> That is a good thing! I would hate for us to get bored.
>
>> 2) I think the implementation was a starting point, but there is much 
>> work to do. Since other contributors (like you) are putting their 
>> time in it, I feel more and more uncomfortable to see only Sanjay's 
>> and my name as authors. I would prefer to have a list of contributors 
>> instead of authors and also to have only geotools' name there, in 
>> order to make it more appealing for further contributions.
>
> Why don't you offer to share "module maintainer" status with Graham? I 
> am really just advising him on this one (and most of the advice has 
> been send more email). He has been reading both the code and the ISO 
> specification for weeks now.
>
>> I'll take a look at the code on this weekend. I have understood the 
>> bug and I will think of it. Meanwhile I have found many things that 
>> should be improved or corrected.
>
> That is good, we would welcome the chance to collaborate here.
>
> Next time we do this email thing can we please send this stuff to 
> geotools-devel (we try to be an open source and an open development 
> project - something that is very hard when we get into the details 
> like this).
>
> Jody



-- 
Graham Davis
Refractions Research Inc.
gd...@re...