Pull request #7: DevelopmentMerge in OI/modelmanagement from development to master
* commit 'ebb03b22d534570011d4592e872e65d72b0fdac3':
Wrong project name
Submodule update
Pull request #5: Fix for finalMerge in ODY_STUDENT/modelmanagement from development to master
* commit 'e4a70a5eceb976774ee9a1b152de8e8df83f438a':
Fix for final
BugfixSmall Bugfix, because not all sql command got the latest entry.
Also small fix on the procedure.
Now everything is tested with multiple container.
Pull request #3: Update based on rechecking the diagram and the codeMerge in ODY_STUDENT/modelmanagement from development to master
* commit '7284969b1de1780c47084e9d2bb13d07898c0007':
Update based on rechecking the diagram and the code
Update based on rechecking the diagram and the code- Updated version of the uml diagram is inserted.
- Some changes on the code, like removing not working stuff and adding some working stuff.
- The commandmanagers now also support a use in a different place, because its not template based.
- Added a new AMUsage as new Part of the AMStoreKey, which indicates if a commandmanager is working in the stream (with a operator) or like already inserted ...
Pull request #2: Bugfix of Bugs that was found during research with the UML Diagram.Merge in ODY_STUDENT/modelmanagement from development to master
* commit '7e4f92981b8b07ed4899f97530470d3da1740e21':
Bugfix of Bugs that was found during research with the UML Diagram.
Bugfix of Bugs that was found during research with the UML Diagram.Fixes:
- In the DB-Script the Procedures use now LONGTEXT as Datatype for the modeldata Parameter.
- As Base for the ParameterStores the interface IAMParameterStore is now used.
- Inside the DB related commands the lock objects are static.
- The variables inside of the AMStoreKey are now final.
- IAMParameterStore got 3 extra methods to make it available that every stores implements some getter...
Pull request #1: DevelopmentMerge in ODY_STUDENT/modelmanagement from development to master
* commit '45259e5e087fd889badef875d92f99e053390643': (22 commits)
Final-Version V1
Added a new Operator, because the CombineAO was not what i wanted. Reworked the Code to work with the new Operator (CombinedPackerAO). Removed some workaround code, because now it is working. ToDo until next and hopefully last commit: -testing o...
Final-Version V1If some exception accour during the testing on a other machine,
a bugfix can be append.
Done:
- Rework / Cleanup of the sourcecode.
- Adding some base comments to the sourcecode.
- Removing not useful methods of the sourcecode.
- Adding a new Operator (CombinedPacker), which allows correct work of MM and EM.
Future:
- Some or call it a really small amount of parts of the code are designed for...
Added a new Operator, because the CombineAO was not what i wanted. Reworked the Code to work with the new Operator (CombinedPackerAO). Removed some workaround code, because now it is working. ToDo until next and hopefully last commit: -testing of the operators (had tested them, but after the last changes, new tests are needed). -updating the sql files. -updating the database structure to use sa...Notes for history:
-A User cant set the model parameters, they are given by the operator (if he
implemented the interface, which is today no one).
-Only an Experiment can given parameters for experimental use.
-Parameters can only changed or set by retraining by a learn_operator like mentioned
in the first history point.
-Some decisions of the design of that system are made by the background, t...
Implemented the Experimentmanagement. 1.)ExperimentManagement_Load Operator is working like it is designed. 2.)Files are renamed after to MM,EM and AM. ToDos. 1.)Testing and maybe fixing ExperimentManagement_Save Operator 2.)The Database should get some same attribute names. 3.)Create some Consts to use for some Values in MM and EM. 4.)Find all parts where the Word "Classifier" is used and rena...Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Rework of the base structure or backend of the experiment and model management. The old model management is moved into that new structure, which is more extendible in a easy way. Also some safetys and defaults are now included. Next Step is to check the moved MM Code for some Parts that need to be renamed like classifier to model and to check if a move of that code to AM as the higher level is ...Note: the structure is build to also allow to extend or use it with systems that
are not a experiment or model management, plus thats the last changes that are
changing the structure.
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Big and last Change of the Modelmanagement. Change: Reworked the full Code-Structure to the same Structure as the last Conception of the Modelmanagement. It is now more scaleable and definitly simpler to extend by new Storages or new Save/Load Commands. Plus the hole structure is already build for some future features ;-) . Everything works like before, Test of load and save worked perfect.Note: This new Structure now is based on a mix of basic and mutated pattern
to support the requirments given to the modelmanagement by me. Also the code
should now be more readable even if some comments are missing.... .
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Modelmanagement is working. ToDo: 1.)Rework or Add Comments. 2.)Add SQL-Code for Creation of Database and Procedures. Fnished: 1.)Testing of Modelmanagement 1.1)Single Save 1.2)Single Load 1.3)Save and Load
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
ModelManagement_Save Operator is now working, but there is much potential.Some steps are needed to finish the basis of the modelmanagement.
First Test was successfully for a new Model (not retrained).
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Added to Classes for the Store-Operation of the Modelmanagement. Implementation is following the next Commits.
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Reworked Load-Operator to use Maps for the Command-Patterns. Changes are testes and are working
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
Update: ~ Added a Logicaloperator for the Load-Process. ~ Moved the Retrieval-Part into the Logicaloperator. ~ Removed the PreTransformation of the Retrieval-Part. ~ The Direct-Call of the Load-Process works like the Retrieval-Part. ~ Each Load-Operator have its own TimerSource and can change the Update-Rate. ~ The Load-Operator is transformed (if the connection is a database) to a DBEnrich-Ope...ToDo bases on this Update:
~ Check the Exceptions.
~ Add Exceptions for bad user inputs (like same uniqueopid).
~ Cleanup of the Code and some comments.
~ Test with a real Classifier that is Base64 encoded inside of the Database (at this time there is some blabla Stuff inside of it)
Rework of the Retrival-Part to remove reflection and use a pattern combination of strategy and command. Some rework of structure of the files inside of the packages and re-adding some retrival methods and remove a last deprecated class is todo stuff.At this stage the parts of retrival are only for retrival, if another section
needs that parts, i will make them more global.
Signed-off-by: Simon Stahmer <Simon.Stahmer1@uni-oldenburg.de>
MM_Retrival Command works, multiple times the command is callable.Only getByIdent* is tested, but rest will be tested soon.
Some Rework of some parts of code is needed to reduce duplicate code.
Working on Retrival-Method. A method call from Odysseus-Script is possible andrestricted to runtim methods and their parameters.
Need more work on the pretransform, in this stage of getByIdentifier is only some testcode
to understand the work of the LogicalPlan static calls.
-Added some code to get nested maps inside of parameters, but need to work on a fix for theprovider because at this time when i look into it, it seems to wont work as it should.