thank you too for your further investigation.
Yep, changing all the methods is a bit of an headache. I haven't found 
out yet (i.e. traced) if the createX() methods are really used to create 
a new FeatureCollection...

But it is good that you found only two algorithms. I forwarded that your 
email to sextante-dev. Lets see what they say... well last time it was 
silence ;)

thanks!
stefan

Am 08.11.12 19:19, schrieb Michaël Michaud:
> Hi Stefan,
>
> Good catch !
>
> It would be fine if the problem could be catched earlier.
> I've not a clear view of what's happening but there are at least two places
> in Sextante binding where a FeatureSchema with a null type can be created :
>
> es.unex.sextante.openjump.core.OpenJUMPTable#create(String,Class[],String[],String)
> and
> es.unex.sextante.openjump.core.OpenJUMPVectorLayer#create(String,Class[],String[],String)
>
> If we patch these method, we must also take care of the attributes
> themselves in
> addRecord(final Object[])
> and addFeature(Geometry,Object[])
>
> Of course, there is still the question of the conversion (to int or to
> String)
>
> The best solution would probably be to eliminate Long.class attribute
> from Sextante
> algorithms.
> I found only two algo which create "Long.class" attributes
> FixedDistanceBufferAlgorithm
> VariableDistanceBufferAlgorithm
>
> The attribute is used to set an id to the features, but there is little
> chance to
> exceed 2^31 features in a Sextante process.
> Moreover, in the schema used for these algorithms, there seems to be a bug
> sFieldNames[0] = "ID";
> sFieldNames[1] = "FROM";
> sFieldNames[1] = "TO";
>
> my 2 cents
>
> Michaël
>
>> Hi,
>>
>> as I wrote earlier I discovered that the fixed buffer tool from Sextante
>> results in an error message with OpenJUMP, when Sextante tries to write
>> a newly created vector dataset as shapefile.
>>
>> Note: When Sextante creates a new layer, it does not add the layer
>> in-memory/directly to OJ, but it writes the data into a shapefile first
>> (using OJs shapefile writer) and then re-loads the shapefiles into OJ.
>>
>> Now I figured that the exception is thrown because Sextante uses the
>> AttributeType long, which OJ does not know.
>> I have modified ShapefileWriter now to see if an attribute has a name in
>> case the AttributeType is null. If it has a name, then the attribute is
>> treated as String. This way we have at least some output from Sextante.
>> But it is not optimal and I wonder if there are more such things?
>>
>> The Long (64) values can not be safely converted to Int (32) values. So
>> it would not make sense to just set the type from Long to Int (in
>> AttributeType).
>>
>> Any ideas? or just leaving as is except my modifications to the
>> shapefile writer?
>> Not sure if it would have any value to implement AttributeType.LONG; I
>> doubt it.
>>
>> cheers,
>> stefan
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Everyone hates slow websites. So do we.
>> Make your web apps faster with AppDynamics
>> Download AppDynamics Lite for free today:
>> http://p.sf.net/sfu/appdyn_d2d_nov
>> _______________________________________________
>> Jump-pilot-devel mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
>>
>>
>
>
> ------------------------------------------------------------------------------
> Everyone hates slow websites. So do we.
> Make your web apps faster with AppDynamics
> Download AppDynamics Lite for free today:
> http://p.sf.net/sfu/appdyn_d2d_nov
> _______________________________________________
> Jump-pilot-devel mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel
>

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_nov
_______________________________________________
Jump-pilot-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jump-pilot-devel

Reply via email to