Delete Request Passing Paramaters about restkit

[DavidBoyes]

I've fixed this on my local copy by editing the method objectLoaderForObject method to check the request type and only include the paramaters if it is not a DELETE request:
if (method != RKRequestMethodDELETE) {
loader.params = params;
}

I believe that the object is supposed to be deleted from the core data store as well, but it appears this isn't happening.

Seems to be an error with this method
- (void)processLoadModelsInBackground:(RKResponse *)response

Seems flawed in how it works, it doesn't seem to reach the bit where it deletes the object as the logic checks for self.targetObject which is set by the objectLoaderForObject method.

Seems like it should be checking for self.method at some stage and then running deleteObject on the targetObject?

[blakewatters]

David -

I have not seen an instance where one would want to include parameters in the DELETE request, but its certainly possible. The right place to fix this for Rails is probably in the RKRailsRouter, where we can tie off the payload specifically for Rails idiomatic requests.

It does look like we should add support for deleting the object under the check for
if (self.targetObject) {
...
code path. The support at the bottom is for removing objects that don't appear in a loaded collection (i.e. I load an authoritative collection of objects and something in the cache is gone, it should be deleted). A check for method == DELETE and a successful response should trigger a delete of the persistent object.

[DavidBoyes]

Ok, I used the dynamic router as it did what I needed so haven't looked at the RKRailsRouter. I hacked on delete with something like this

if(_targetObjectID && self.method == RKRequestMethodDELETE) {
    NSManagedObject* backgroundThreadModel = [self.managedObjectStore objectWithID:_targetObjectID];
    [[objectStore managedObjectContext] deleteObject:backgroundThreadModel];
} 

Which worked but was mainly to test it out. I've had to swap to my own httpriot based model manager because I couldn't get certain things to work, but am keen to see RestKit progress so might contribute when I can as my own rolled one doesn't use coredata.

[blakewatters]

David -

We are adding support for the use-cases you outlined. What other issues have you been running into preventing adoption of RestKit? We are happy to fix/enhance the framework or offer support if it helps you use (and later contribute to :-) the library.

• Blake

[DavidBoyes]

Posting with data was just flat out not working, my Rails dev server wasn't even acknowledging that it had been hit. It was pretty late at night so I sort of gave up without much debugging. I can swap to RestKit relatively easy so when I realise that adding my own local caching will be to much extra work I will probably swap back.

[blakewatters]

We have just pushed a fix that adds deletion of individual objects when a DELETE request is successful. We've also updated the Rails router so that it returns a nil serialization for DELETE requests against objects.

Swap it back when you have a change and we can see if things are working better.

Blake

[DavidBoyes]

I will swap it over and try it out.

[blakewatters]

Any update here?

[DavidBoyes]

Sorry, it appears to be fine now. Thanks