The travis CI builds are currently failing for openjdk9 and openjdk10. The builds should be fixed and both jdk9 and 10 should be added to the .travis.yml.

http://www.mapbox.com/developers/utfgrid/

Forcing the user to specify the trailing slash.

GeoTools is coordinating an update to JTS 1.15 for our next release cycle and would like to ask Jeo to take part.

For details see: https://github.com/geotools/geotools/wiki/Upgrade-to-JTS-1.15.0

The project has changed package names from com.vividsolutions.jts to org.locationtech.jts, update instructions are available.

Just walking through some code, I noticed what appears to be a duplicate delegate cursor:

https://github.com/jeo/jeo/blob/master/format/geopkg/src/main/java/org/jeo/geopkg/GeoPkgWorkspace.java#L367

From what I can see, the call to QueryPlan.apply adds this. I couldn't see any failures or, oddly, great performance differences but I guess with larger datasets it might appear.

This also appears in jeo-android.

At the moment all forms are written so that they include all attributes from the underlying format. It is much more efficient to include only those attributes that are required.

Long, Short, etc. -> INTEGER (from sqlite docs: The value is a signed integer, stored in 1, 2, 3, 4, 6, or 8 bytes depending on the magnitude of the value.)
Float -> REAL

This will require a better (or reversed) data structure than used in GeoPkgTypes (as it only allows a single key).

These merit discussion:
BigDecimal -> ?
BigDecimal comes from the shapefile driver currently. When writing to a gpkg, the field gets transformed into a String (TEXT) due to a missing mapping. The issue is the potential for data loss if we use a different storage type. Remaining as a string may not work either.

Date/Time -> ?
As there is no date support in sqlite so looks like we have to do this manually. While I can create a column as CREATE TABLE Y(value DATETIME); and see that the schema records this, the JDBC driver doesn't return this in the DB metadata, making our task more difficult.

In package io.jeo.filter.cql the following files are missing:

ParseException
ECQLParser
FilterCompilationUnit

the parsers can throw TokenMgrError (not exception subclass). This should be caught and translated.

One of the requirements of pushing to central is that the groupId match a domain that the project has control over. We have jeo.io but not jeo.org.

Happens when you have a .jeo meta file with a basename that matches another file in the same directory.

As described in #92. You can do this with iterators in java which I think is a strong argument for it.

or the map will not prevent zooming in/out beyond the data resulting in pink tiles

The latest version of jeo I see in maven central is v0.6 is v0.7 available in a different maven repo?

I have downloaded the jeo-master zip file. Having installed maven I have run this command:
mvn install

this produces the following error, which I also get having loaded jeo-master into Netbeans, and issued a build command.
Please could you help suggest what I am doing wrong?
Thanks.

T E S T S

Error: Could not find or load main class Files
Picked up JAVA_TOOL_OPTIONS: -Djava.vendor="Sun Microsystems Inc."

Results :

Tests run: 0, Failures: 0, Errors: 0, Skipped: 0

[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary:
[INFO]
[INFO] jeo ................................................ SUCCESS [ 4.434 s]
[INFO] Core ............................................... FAILURE [ 8.876 s]

patch done already

To encapsulate write operations to Feature. This would make it easier for format drivers to optimize for read only cases.

Painful mass change but best to be consistent with the domain name of the project. This is also a requirement of deploying to maven central.

To allow the format to evolve and remain backward compatible, or at least provide a sensible error message include a version in the format.

Add a GeoTIFF driver.

This is due to autocommit.

Adding support for reading shapefiles.

This appears in several places but some examples:

http://192.168.0.15:8000/features/va/populated_places.html?limit=1&srs=epsg:9999
org.osgeo.proj4j.UnknownAuthorityCodeException

http://192.168.0.15:8000/features/va/populated_places.html?limit=1&srs=9999
org.osgeo.proj4j.UnsupportedParameterException

http://192.168.0.15:8000/features/va/populated_places?limit=1&srs=xyz:9999
java.lang.IllegalStateException: Unable to access CRS file: /nad/xyz

Looks like geobuf from mapbox has some steam behind it, no sense in maintaining a custom protocol buffer based format.

it should handle throwable

The classes are identical, might as well factor them as there is already an sql util module.

The current implementation is slightly out of date. The acceptance test should be the [sample data](http://www.geopackage.org/#sampledata from the geopackage site. We should also ensure we can read geopackages created by ogr.

Feature f = new MapFeature("id");
Geometry g;
f.put(g);

Under this case, g will not be put to the map anyway.

STATE_NAME IN ('Virginia')

for example

Currently the json structure we use for objects like datasets are duplicated. Factor them out into the core module so they can be re-used.

support retrieving a single feature and editing one

@jdeolive could you please tag and push a version 0.8?
https://jitpack.io/#com.github.jeo.jeo/jeo currently shows 0.7 as newest version which is quite far in the past. With the push of a new release tag, jitpack would also list the newest version and other projects could refer to 0.8 instead of master-SNAPSHOT.

Currently, the mongo tests are failing in Travis CI and the module was removed in 114440f. For more details on the failure see https://travis-ci.org/github/jeo/jeo/jobs/676843668

The tests should be fixed and the module should be added to the build again.