neo4j-contrib
index on wkt property not removed cleanly
(http://groups.google.com/group/neo4j/browse_thread/thread/630dc8ce8c87cf06)
it seems when deleteing a node containing a wkt property the wkt property is not removed from the index, even the node is cleanly removed from the DB. This might result in a NotFoundException in case of spatial search execution as nodes will be returned which actually are no more present in the DB.
Workaround/hack: catch NotFoundException and treat this as a common bahavior..
cheers, Florian;
Florian, do you think you could write a failing test for it and maybe fix it? Would be appreciated!
I am also seeing similar issues on deleting nodes with wkt property. Is there any new development on this? Thanks. Eugen.
Not yet, would love for you guys to add a failing test with a PR, maybe it is easy to fix?
I will scrub my current test from what's not needed and create a gist out of it just as soon as I can get this other issue working: #39 Thanks for the quick feedback. Eugen.
Grgeat, you can add it to the test package, and send in the CLA according to http://docs.neo4j.org/chunked/snapshot/cla.html so I can merge it directly :)
/peter
On Sat, Mar 17, 2012 at 3:43 PM, Eugen [email protected] wrote:
I will scrub my current test from what's not needed and create a gist out of it just as soon as I can get this other issue working: #39 Thanks for the quick feedback. Eugen.
Reply to this email directly or view it on GitHub: https://github.com/neo4j/spatial/issues/36#issuecomment-4554409
Hi, just wondering if anyone has tried to tackle this yet? I can put together a test case but it's very simple to see in action. I created a single node with wkt set to 0,0. Fired up Neoclipse and saw the spatial index node and the real node. I deleted the real node and the spatial node continued to exist with an id value of the node that I just removed.
Hi,
I think I have a fix for this, but could do with you guys having a look to make sure I'm not doing anything untoward. I'm currently working on a fork of spring-data-neo4j and spatial, but was unsure if I should just make changes to master and make a pull request, or make a branch. Any preference?
Cheers,
Ben
Since your fork is already implicitly another branch, I personally see no reason to make yet another branch. Just fix on your master and send the pull request.
I think you should only consider making addition branches in your local repo if you were two multiple parallel types of work you wanted to keep separate for your own work, or perhaps you were tracking multiple branches in the origin. I don't believe either are the case here.
Sorry, clicked 'close and comment' instead of just 'comment' before. Slip of the fingers :-)
Nice one, cheers Craig. Once I get home I'll get the changes checked in and send the pull request.
Damn, looks like I've broken something unintentionally, org.neo4j.gis.spatial.IndexProviderTest.testNodeIndex() seems to be failing on the last assertion, expected 2 but getting 0.
I'll try again tomorrow.
I may have to check it in with the failing test and get you to have a look. Poking around trying to fix this has made me realise just how little I understand about what is going on under the hood :-/
Hi Craig,
I think I've got something worth looking at now.
I've raised a pull request for both spatial and spring-data-neo4j, as listed below.
https://github.com/neo4j/spatial/pull/67 https://github.com/SpringSource/spring-data-neo4j/pull/90
Both projects build successfully, and I've added an additional test to prove the expected behaviour.
I'll look forward to your feedback.
Cheers,
Ben