Unsuccessful deployments show up as successful when deploying from modeler

[benhoffmann]

If I deploy a model directly from the modeler, it says "Deployment was done successfully" even if this was not the case. My engine complained with a BPMN parse error and did not deploy the diagram, but the modeler did not catch that problem.

v 1.16.0

[fractallian]

Would be extremely helpful to surface model parsing errors in the modeler. I see errors using the API but modeler just says deployment succeeds for the same file.

[Alexander-Kiselyov]

Create Deployment endpoint does respond with an error - HTTP 500 - when I try to deploy a definition with a process-level Event Subprocess triggered by a Compensation Start Event (not supported according to the doc):

> POST /engine-rest/deployment/create HTTP/1.1
> Host: localhost:8081
> User-Agent: curl/7.55.1
> Accept: */*
> Content-Length: 12747
> Expect: 100-continue
> Content-Type: multipart/form-data; boundary=------------------------75a62bae9b8408af
>
< HTTP/1.1 100
< HTTP/1.1 500
< Content-Type: application/json
< Transfer-Encoding: chunked
< Date: Sun, 23 Sep 2018 20:24:26 GMT
< Connection: close
<
{"type":"ProcessEngineException","message":"ENGINE-09005 Could not parse BPMN process. Errors: \n* event subprocess with compensation start event is only supported for embedded subprocess (since throwing compensation through a call activity-induced process hierarchy is not supported) | testprocess.bpmn | line 30 | column 48"}* Closing connection 0

I'm not a JS expert, but as I understand, deployment logic is located in app/lib/createDeployer.js, and, as far as I see in fetch() docs, 500 won't lead to throwing an error.

[nikku]

Closed via #1041.