[postgis-devel] strategy about empty

[Nicklas Avén]

Sorry, just the half thought in last post... By strategy I mean we have to handle the discussions on how it should behave, and to get so consistent behavior as possible between didderent functions. We also have to register in a straight way where er are today. How it is handled today and in earlier versions. For example in the case of st_distance it is now reutning null as proposed in DevWikiEmptyGeometry instead of 0 as before. For someone this might be a breaking change. In st_dwithin it returns false as said in the wiki-page above but there is also a discussion about if it instead should retun null. Where to have such a discussion so we see how one decission affect consistently for other functions. Should it all be in the wiki or mail?  /Nicklas  
2009-11-28 Nicklas Avén wrote:

>
I have looked a little at this #294 thing about empty geometries cases>
 >
I think some strategy is needed to manage that one because it is big and will need a lot of different discussions I can imagine.>
I have looked through a lot of the functions of today, what they give with empty geometries and found some inconsistencies for example:>
select ST_NumPoints('POLYGON EMPTY')  returns nothing, while>
select ST_NPoints('POLYGON EMPTY') returns 0
>>
I think we in the end will need a matrix with the behavior in different cases with different versions. >
Should that be a part of documentation or wiki?>
 >
/Nicklas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20091128/a7a73a01/attachment.html>