Upload
ordell
View
33
Download
1
Embed Size (px)
DESCRIPTION
Making TTCN-3 work!. Issues and strategies for its use in product development. Martin Botteck, Thomas Deiß, Colin Willcock Nokia Research Center Bochum, Germany. TTCN-3 adoption today:. Generally, uptake is slower than anticipated. Speed-up could be possible through : - PowerPoint PPT Presentation
Citation preview
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:1
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Making TTCN-3 work!
Issues and strategies for its use in product development
Martin Botteck, Thomas Deiß, Colin Willcock
Nokia Research Center
Bochum, Germany
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:2
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
TTCN-3 adoption today:
• Generally, uptake is slower than anticipated. Speed-up could be possible through :
• Handling variations of the testing task
• Integrating TTCN-3 into existing test environments
• Using TTCN-3 in the hardware development process
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:3
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Overview
TTCN-3 for testing hardware
Conclusion/Summary
TTCN-3 in existing test environments
Variations in Test Systems
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:4
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Variations in Testing
• Within a set of test cases there often are several commonalities
• Example: 3G “Call”
– several types of calls (voice, data,…) and calling routes– behaviour will in many cases be similar, but not the
same– “intuitive” solution: copy/paste code from one test case
to another “similar” one• -> maintenance problem when the 3G specification is
updated– alternative solution: add a parameter in the test case
invocation• -> code will become hard to read
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:5
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Types of variations
• Value Variants– Variations that may be expressed by altering a
parameter inside a test case or message sent to the SUT
– seems trivial but soon becomes difficult due to • vast amount of parameters and combinations• typically not all combinations make sense• dependancies between the values is typically not
documented and implicit– a language or coding style solution needs to be
found for this
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:6
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Types of variations (ctd)
• Behavioural Variants– Variations that may be expressed by a different
sequence of messages between the SUT and the tester• several messages or sets of sequences may remain the
same• some are different
– decomposition of the SUT may reveal common abstract test case structures with respect to the SUT components
– but what about complex composition variants of such components?
– -> Abstract Modelling of the tester? (UML testing profile, Test Purpose Language)
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:7
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Types of variations (ctd)
• Variants of the System under test– product customisation– evolution of conformance standards
• Might be expressed by altering a parameter (e.g. “feature_x=OFF”, “call_type=‘rich_call’”)
• in combination with reference to a set of message sequences
– Typically, a vast amount of such combinations will be needed
• How to solve this without copy/paste of code but still avoiding combinatorial logic to the user?
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:8
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Overview
TTCN-3 for testing hardware
Conclusion/Summary
TTCN-3 in existing test environments
Variations in Test Systems
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:9
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
The dilemma of choice
• Nowadays most R&D departments already perform testing:
• Conformance testing, hardware testing, product testing, module testing, unit testing…
• Proprietary test systems (“home grown” or “tailor made”)
• Multitude of vendors with closed environments
• Trying out a system is a lot of effort. So, shortcomings of any choice being made show up too late for reversing the choice
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:10
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
The testing tool chain
• for each stage components from different vendors need to interoperate
• clearly defined interfaces are needed
Testware
Test cases
Test System
Codec
Router
Test Development Tools
Editor
Compiler
Codec Generator
…
Test ExecutionControl
Log
Analysis
Thomas Deiß:
Usually, in the TTCN-3 world, the platform and the SUT adapter are
considered to work side by side, not the PA on top of the SA.
MBo: Better this way?
It does not become clear to me, what are the stages: Tools, Testware (development), Test sytem
(development?), Test execution. But 'tools' is not a stage.
MBo: Indeed. Let´s talk
Editorial comment: Test system <-> Test Execution, note different capitalization of
the second word
Thomas Deiß:
Usually, in the TTCN-3 world, the platform and the SUT adapter are
considered to work side by side, not the PA on top of the SA.
MBo: Better this way?
It does not become clear to me, what are the stages: Tools, Testware (development), Test sytem
(development?), Test execution. But 'tools' is not a stage.
MBo: Indeed. Let´s talk
Editorial comment: Test system <-> Test Execution, note different capitalization of
the second word
Run Time System
Executable Test System (ETS)
PlatformAdapter
SUTAdapter
SUT
Codec
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:11
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Run Time System
Executable Test System (ETS)
PlatformAdapter
SUTAdapter
SUT
Codec
The testing tool chain interfaces
Testware
Test cases
Test System
Codec
Router
Test Development Tools
Editor
Compiler
Codec Generator
…
Test ExecutionControl
Log
Analysis
• TTCN-3 core notation language for exchange of test cases
• Run Time IF for Adaptors and Codecs
• Test Execution Control and logging through TCI
TCI/LogTRI
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:12
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Further work on interfaces
• Tool vendors have a notion to “close shop” and not support public interfaces
• Codec interface part in TCI leads to non-interchangeable codecs
• realisation of a given specification in IDL depends on the target language.
• Java based tools and C based tools require different codecs• performance problems when separating the codec to a
separate executable• Standardisation sometimes lags behind (e.g. TCI)
• It is also desirable to generate as many test cases and parts of the environment as possible directly from the behavioural description of the IUT
• -> modelling representations (e.g. XMI) need to be supported in the tools
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:13
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Overview
TTCN-3 for testing hardware
Conclusion/Summary
TTCN-3 in existing test environments
Variations in Test Systems
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:14
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Hw design flowAlgorithmic description
informal model, e.g. C source code
Highest Level behavioural model
(e.g. SystemC)
Decomposition
(Block diagram)
Notion of “Time”
Bit accurate Hardware model
(e.g. VHDL)Notion of “Clock Cycles”
Physical Model
(e.g. RTL, Spice,…)
Introduction of physics
(layout, transistor, …)
• Description consistent through all stages/levels
• Automatic generation of tests (“Test vectors”)
• Back annotation integral part of the process
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:15
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Sw design flowuse case description
formal model, e.g. UML UC
Highest Level behavioural model
(e.g. UML sequence diagram, state chart)
Decomposition
(package diagram)
Refinement,
Introduction of “Signals”
executable
(e.g. binary, script,…)
Introduction of programming language syntax
(C, C++, Java,…)
more detailed behavioural model
(e.g. UML sequence diagram, state chart)
• Rather “novel” approach• Automatic generation of tests (“Test vectors”) still in experimental phase• Back annotation is not formally specified
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:16
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
“Traditional” HW-SW integration
Algorithmic description
informal model, e.g. C source code
Highest Level behavioural model
(e.g. SystemC)
Decomposition
(Block diagram)
Notion of “Time”
Bit accurate Hardware model
(e.g. VHDL)Notion of “Clock Cycles”
Physical Model
(e.g. RTL, Spice,…)
Introduction of physics
(layout, transistor, …)
use case description
formal model, e.g. UML UC
Highest Level behavioural model
(e.g. UML sequence diagram, state chart)
Decomposition
(package diagram)
Refinement,
Introduction of “Signals”
executable
(e.g. binary, script,…)
Introduction of programming language syntax
(C, C++, Java,…)
more detailed behavioural model
(e.g. UML sequence diagram, state chart)
SW release HW Prototype
SW release HW prototype
Integration Integration Tests
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:17
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Common nodes in the HW and SW design flow
• Integration tests are not determined by formal derivation from HW/SW high level models
• Integration typically is the 1st stage where HW and SW meet
• Integration results feed back “somewhere” in the process in SW
• HW and SW tests only derive from HW and SW models respectively
• How about
• utilisation of SW models (e.g. High level behavioural) for generation of HW tests?
• including behavioural HW models (SystemC) in High level behavioural SW models?
• generation of tests directly from the model representation?
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:18
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Overview
TTCN-3 for testing hardware
Conclusion/Summary
TTCN-3 in existing test environments
Variations in Test Systems
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:19
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Summary/Conclusions
• Further deployment of TTCN-3 needs convincing advantages:– integration into existing test environments
• interfaces, replaceable components from heterogenous test systems
– handling of variants through specific coding constructs
– automated generation of test cases from abstract models
– cover HW as well as SW through adequate representation in SW models
© NOKIA Originator: Martin Botteck / April 12, 2005 / Page:20
Nokia Research Center
CA
R/M
EM
/VTT
Making TTCN-3 work
Thank you! Questions?
Acknowledgement: The authors would like to thank Mr. Risto Teittinen from Nokia Networks for his contributions to this topic