New versions of Adlink Edge SDK are released on a regular basis. This page lists all the known issues for Adlink Edge SDK V1
You may also want to read the following:
Edge SDK Known Issues
| Report ID. | Description |
|---|---|
| THINGAPI-1037 | FlowIdFilters not working on Purged samples A read(), read_next() or data-available listener may return purged samples with flow id's that do not match the configured 'flowIdFilters' property of the Thing's input. Workaround: You must filter such purged samples manually. Each such sample will have the following characteristics: a) the flow-state will be PURGED; b) if the sample is IOT_NVP_SEQ data, there will be no values in the returned sequence; c) if the sample is PROTOBUF data, then VDataSample::hasData() will return false. |
| THINGAPI-623 | A Python Listener class defining a constructor may cause a Segmentation Fault/Access Violation A Python Listener class that defines a constructor (that is, a __init__ method) must call the super class's __init__ method, otherwise a Segmentation Fault/Access Violation will occur when a listener instance is added via a call to one of the Python API's add_listener methods. Workaround: See the Python API User's Guide for an example of a well formed listener class with a constructor. |
| THINGAPI-438 | Control-C not received by Python applications during blocking read/read_next During a blocking call to Thing.readIotNvp, Thing.readNextIotNvp, Selector.readIotNvp and Selector.readNextIotNvp, any Ctrl-C interrupt will not be received by the Python engine. |
| THINGAPI-533 | Excessive cpu-load for ThingAPI read-operations on large (e.g. multi-megabyte video-)samples When applicable: if cpu-monitoring tools such as 'top' shows about equal amount of system-time and user-time for read-operations Work-around: to reduce the cpu-load create 2 environment variables with appropriate values: export MALLOC_TRIM_THRESHOLD_=-1 (which causes 'mmap' to be used rather than malloc for large memory-chunks export MALLOC_MMAP_THRESHOLD_=4000000 (a value greater than the sample/video-size) Consequence: increased memory-usage of a process which decreases the amount of available RAM for all other processes |
| THINGAPI-766 | Python API treats IOT_INT8 as a Python string In Python, a data type of IOT_INT8 gets incorrectly converted into a python String. It is intended to be an integer value with range -256 through 255. |
| THINGAPI-880 | NodeJS API is incomplete The NodeJS API shipped with earlier releases of EdgeSDK remains incomplete. In particular: * the NodeJS API contains no support for listeners. * Examples that rely on listeners are not available. * No User's Guide is available. |
| THINGAPI-933 | Protobuf examples without 'CMAKE_BUILD_TYPE=Release' on Windows Protobuf Examples configured on Windows without '-D CMAKE_BUILD_TYPE=Release' appear to work. However during a build, errors appear: "error LNK2038: mismatch detected for 'RuntimeLibrary': value 'MD_DynamicRelease' doesn't match value 'MDd_DynamicDebug'". Work around: Add "-D CMAKE_BUILD_TYPE=Release" to the cmake command. |
| THINGAPI-934 | Limitation: Pre-1.4 applications cannot fully inspect Protobuf TagGroups Applications such as the 'thingbrowser' application inspect all TagGroups found on a DataRiver, and provide information on their contents by making use of TagGroup methods such as getToplevelType(), getTopLevelTypename(), getTypeDefinitions() and getType(). Pre-1.4 versions of such applications will 'see' Protobuf TagGroups, but they will not be able to find information their content. Migration to 1.4: When such applications are migrated to 1.4, developers must examine their use of the above methods. On Protobuf TagGroups, they will now throw a C++ exception. Migrated code should first examine the result of the new method 'TagGroup::getWireFormat()' to determine whether the Tag Group is the 'familiar' NVP_SEQ-format or the new PROTOBUF-format. New TagGroup methods are provided in 1.4 to examine the contents of a protobuf tag group. See the C++ User's Guide for details. |
| THINGAPI-937 | Windows: C++ Examples cannot use the default Visual Studio 'generator' On Windows, the C++ examples cannot be configured without explicitly specifying -G "NMake Makefiles" on the command line. Without this option, CMake attempts to use the default Visual Studio generator. However, some of the C++ examples cannot be configured this way. Work-around: add the option -G "NMake Makefiles" to the CMake command line, and then use nmake to build. This process is described in the examples README file. |
| THINGAPI-940 | Windows+Protobuf: C4275 and C4800 Warnings in header files EgdeSDK Protobuf applications compile with a C4275 and C4800 warnings. These can safely be ignored. Google documents that C4800 is unavoidable ADLINK is researching how C4275 can be addressed. Work-around: These warnings can safely be ignored. |
Operating System/Platform Related Issues
- As of ADLINK Edge SDK version 1.0.1, Ubuntu 14.04 is not supported.
- As of ADLINK Edge SDK version 1.1, Ubuntu 18.04 support has been added.
- As of ADLINK Edge SDK version 1.3, Ubuntu 18.04 AArch64 support has been added.
- As of ADLINK Edge SDK version 1.4, Python 3.8 on Linux has been added.