Bugzilla – Bug 321825
[GMCS] Marshalling error 'unknown type 0x13 in type_to_ldind'
Last modified: 2010-04-08 02:48:09 UTC
---- Reported by steveb@mindtouch.com 2006-08-20 02:19:48 MST ---- First, you need the MindTouch Dream source at: svn://dev.opengarden.org/svn/trunk/product/dream To build, execute build.sh Description of Problem: while trying to circumvent https://bugzilla.novell.com/show_bug.cgi?id=MONO79112 by copying the input stream to a MemoryStream, we're running into a marshaling issue. Steps to reproduce the problem: 0. edit 'mindtouch.core/httptransport.cs' and uncomment '#define COPY_INPUT_STREAM' 1. build Dream using 'build.sh' 2. change to the 'bin' folder 3. execute 'mono --debug mindtouch.host.exe' 4. open another terminal window 5. go to the 'mindtouch.host' folder 6. execute 'curl -H 'Content-Type: text/xml' -d @show-headers.startup.xml http://localhost:8081/host/execute' Actual Results: The request will be processed and will throw an internal exception (that's fine, internal exceptions are converted to messages automatically). However a few secoonds later it aborts with: ** ERROR **: unknown type 0x13 in type_to_ldind aborting... Expected Results: Exception should be sent back to 'curl' as an xml document. How often does this happen? Always on OS X. Additional Information: In addtion to the steps of #79112, the following happens: 1. 'HostService.PostStartService' is invoked by DreamEnvironment 2. HostService then attempts to retrieve the blueprint for the service using a DirectPlug (since it's talking to a co-hosted service) 3. 'Plug.Get' fails and throws an exception 4. this exception should propagate to any other waiting operations, but instead results in an internal mono error ---- Additional Comments From vargaz@gmail.com 2006-08-20 07:33:31 MST ---- Please attach some kind of simplified/self-contained test case. Also, this is most likely a generics bug, since 0x13 == MONO_TYPE_VAR. ---- Additional Comments From miguel@ximian.com 2006-08-20 18:07:32 MST ---- With my bug fix to the HttpListener stuff, I can no longer see this, so it must be some code path that is no longer being triggered. Can you guys please make a test case? Unknown bug field "cf_op_sys_details" encountered while moving bug <cf_op_sys_details>OS X 10.4.7 Intel</cf_op_sys_details>
NEEDINFO for 4 years.