[
Date Prev][
Date Next][
Thread Prev][
Thread Next][
Date Index][
Thread Index]
[
List Home]
|
Re: [geomesa-users] java.lang.IllegalArgumentException: Cannot meaningfully index a feature with a NULL geometry
|
Yeah, I believe we try to ensure that either the entire feature is
written, or none of it - so if it has an invalid geometry it wouldn't be
attribute indexed either. As far as a default geometry that wouldn't
show up in any queries, I don't know of anything like that...
I looked back at how this was handled in 1.2.7, and I think that it
should have been pretty much the same. The exception you're getting is
actually not from the indexing itself, but from generating a feature ID,
which changed in 1.3.x. I'm curious if maybe the features haven't been
getting indexed all along, but were just failing silently or being
caught and logged.
To get past your current exception, you could try setting the feature ID
yourself (along with the USE_PROVIDED_FID hint), or changing the feature
ID generator as outlined here:
http://www.geomesa.org/documentation/user/datastores/runtime_config.html#geomesa-feature-id-generator
The default uses a Z index to try to group features close to each other
in the records table, but you could instead use our other default
implementation which is based on ingest time:
org.locationtech.geomesa.utils.uuid.IngestTimeFeatureIdGenerator
Thanks,
Emilio
On 03/13/2017 10:46 AM, David Boyd wrote:
Emilio:
Thanks.
Yes, we are using attribute indexes for non-spatial search.
If I do, the try/catch approach, I assume my non-spatial data will
still not be stored correct?
On putting in a place holder the question would be what value to
put. I think this particular condition is
happening because some records do not have location information. Is
there a standard value for
a Null or empty point geometry (that will either never match or always
match a spatial query)?
If this were an area style geometry, one could do the whole globe to
always match, what could be done to never match?
On 3/13/17 9:29 AM, Emilio Lahr-Vivaz wrote:
Hi David,
We did change our indexing to explicitly reject features that can't
be indexed appropriately. Are you using attribute indices to search
for non-spatial records?
We don't currently have a setting to not fail on invalid data -
generally you would wrap each call to `featureWriter.write()` in a
try/catch and handle the exceptions. Could you put in a placeholder
geometry?
I'll add a ticket to allow for this behavior again.
Thanks,
Emilio
On 03/10/2017 10:38 PM, David Boyd wrote:
All:
Sorry for the dumb questions.
Is this error new behavior in 1.3.X?
We were not getting this (to our knowledge) in 1.2.7.X.
java.lang.IllegalArgumentException: Cannot meaningfully index a
feature with a NULL geometry
Since I am using geomesa to support search of all my data in
accumulo, I need to be able to store records
that have empty geospatial fields.
Is there some setting that needs tweaked?
_______________________________________________
geomesa-users mailing list
geomesa-users@xxxxxxxxxxxxxxxx
To change your delivery options, retrieve your password, or
unsubscribe from this list, visit
https://dev.locationtech.org/mailman/listinfo/geomesa-users
