<html>
<head>
<title></title>
<meta name="GENERATOR" content="MSHTML 8.00.6001.18852"></meta>
</head>
<body>
<div align="left">Ok, now I have committed to spike. </div>
<div align="left"> </div>
<div align="left">Since last review Paul I have changed:</div>
<div align="left">* removed subgeometry handling. Hope I can get back with that one to another release</div>
<div align="left"> </div>
<div align="left">* moved prototypes and structures to measures.h</div>
<div align="left"> </div>
<div align="left">* moved making of line in shortest and longest line to measures.c instead of postgis_functions_basic. Now the structure DISTPTS is only used inside measures.c</div>
<div align="left"> </div>
<div align="left">* changed from void functions to returning integer. true for sucess and false for error</div>
<div align="left"> </div>
<div align="left">* changed to handling LWGEOM instead of uchar. Since I removed the subgeometry handling it wasn't that big difference. Most of the uchar things was happening in the bbox-comparing.</div>
<div align="left"> </div>
<div align="left">I have done some testing and so far everything has passed.</div>
<div align="left"> Regina I tested aginst your dataset and now it seems ok.</div>
<div align="left"> </div>
<div align="left">Things left to do:</div>
<div align="left">handling empty geometries. I'm not sure what the best way of doung that is, because we have to investigate collections before we know if all content is empty or if there is any geometry inside. And what does the empty geometry looks like when it has passed the parser. What is the best way to detect them?</div>
<div align="left"> </div>
<div align="left">the most ugly function now is lw_dist2d_recursive.</div>
<div align="left">I haven't figured out the best way of handling, when iterating the lwcollections, that I don't know if the next geometry showing up has subgeometries. Now I check for that several times. think the code works, but it looks very messy.</div>
<div align="left"> </div>
<div align="left">Please review it again</div>
<div align="left">I don't think I will manage to build the unit-tests in a smart way in time so I hope someone else can do that. </div>
<div align="left"> </div>
<div align="left">Thanks</div>
<div align="left">Nicklas</div>
<div align="left"> </div>
<div align="left"><br />
2009-11-18 Nicklas Avén wrote:<br />
<br />
></div>
<div align="left">Hallo</div>>
<div align="left"> </div>>
<div align="left">I have ran into troubles I can't handle with the code, or at least it will need time I don't have. Regina also found a bug in the sub-geometry handling which I don't see what it is, at once (shouldn't be to difficult to find though)</div>>
<div align="left"> </div>>
<div align="left">Because of this I think I better let out this subgeometry handling for this version and focus on the faster algoritm and new functions. </div>>
<div align="left"> </div>>
<div align="left">If I remove this "preprocessing" with subgeometries and just implement those suggestions you gave Paul I think I can get it working in a day or two and maybe after testing it can go to trunk after weekend, when I will get some time to work on it. Then I hope to get the subgeometry handling working to some later release.</div>>
<div align="left"> </div>>
<div align="left">Is that ok?</div>>
<div align="left"> </div>>
<div align="left">The problem I have in the code I think is very strange. I cannot see the logic in the crashes. If I for instance force an iteration to only iterate once it all works, but if I let it go a second round it doesn't even complete the first round.</div>>
<div align="left"> </div>>
<div align="left">I have had several situations where I do an edit late in the code which causes a crash early in the code. That makes it very hard to see what's happening. I guess I have made some fundamental thing totaly wrong in the bginning of the code causing it all to be very instable. </div>>
<div align="left"> </div>>
<div align="left">/Nicklas</div>
<p align="left"><br />
><br />
> 2009-11-18 Paul Ramsey wrote:<br />
><br />
> Niklas, Mark,<br />
> ><br />
> >The two largest items left on the 1.5 docket are the new distance work<br />
> >and the upgrades to shp2pgsql. Can you give me ETAs on those? I'd like<br />
> >to start the 1.5 train towards the station...<br />
> ><br />
> >P.<br />
> >_______________________________________________<br />
> >postgis-devel mailing list<br />
> >postgis-devel@postgis.refractions.net<br />
> >postgis.refractions.net/mailman/listinfo/postgis-devel<br />
> ><br />
> ></p>
</body>
</html>