- Also major cleanup of the filter.
- Implementation clompleted for pctl.
Next up: Wrap up the Csl and Ltl filter and then testing.
Former-commit-id: 8189f8462c
- Lots of debugging
- Changed the way the filter keeps information about the scheduler to use for probability/reward queries.
| This was done by keeping a special action at the first position of the action list.
| Which was not exactly consistent with the idea behind the filter actions.
| Now the filter keeps this information as an enum value in a member variable.
- All but one tests are green. So we almost reestablished full functionality.
|- The last test that still fails is SparseMdpPrctlModelCheckerTest.Dice where the second to last model check returns the wrong result.
Next up: Debug. Then introduce the full range of filter actions.
Former-commit-id: fd311966cc
- Also fixed up control flow and some tests for new interfaces.
|-> It now compiles again.
Next up: More functionallity in the filter.
Former-commit-id: 21d43e75c4
- This change splits the path formulas into probabilistic path formulas like Next or Until and reward path formulas like InstantaneousReward or SteadyStateReward.
|- That way it is assured at compile time that no reward path formula can ever be subformula of any probabilistic bound operator and vise versa.
Next up: Adopt changes in the Csl formula tree to the Csl parser.
Former-commit-id: d74c88bbf8
- Every PRCTL formula that worked before works now and behaves in the same way. One exception:
|- Formulas of the type Pmin<0.5[Phi] and Rmin<0.5[Psi] result in a parsing error, as the comparison operator already implies the scheduler to be used.
| Also, the modelchecker now actually uses the comparison operator in order to choose the correct scheduler for MDPs.
Former-commit-id: d942d18e7e
- First one concerning the MappedFileTest in which I neglected to consider that the number of characters used to signal a new line differs between Linux (\n -> 1) and Windows (\r\n -> 2) which caused the test to fail on all OS using two characters (hence not on Linux, where I ran the tests).
- Second bug concerned the case that a transition reward file contained more states than the corresponding transition file.
In that case the parser tried to acces the entry of the rowGroupIndices vector behind the last actual entry, which caused an exception to be thrown.
Now there is a check whether the highest state index found by the parser does exceed the highest state index of the model.
Former-commit-id: bc83267f3c
- Fixed comments.
- It seems to be ASSERT_EQ(expected, actual);
|-> Switched arguments of nearly all ASSERT_EQs to correctly use this macro in the parser tests.
Former-commit-id: e5059709f2
- Some renaming (among others unmatched -> mismatched).
- Added checks and tests for doubled or skipped lines as well as lines concerning the same transition.
Next up: Remerge.)
Former-commit-id: 05efcbf91c
- Done with the tests. At least for now.
|- There are tests for all parsers and helper classes now.
Next up: Some minor fixes and finally the merge.
Former-commit-id: ebb2ea50d5
- Eliminated the need for SupportedLineEndings (I hope...should work on all os, but have only tested this on linux)
- Moved fileExistsAndIsReadable to MappedFile
- Removed scanForModelHint and replaced it with its contents at the one point where is was used in the AutoParser
- Moved and renamed remaining part of Parser.cpp/.h to src/utility/cstring.cpp/.h
|- New namespace of the cstring manipulation functions is storm::utility::cstring
|- Used a using namespace storm::utility::cstring to eliminate the full namespace specification within the needing parsers. This keeps the code readable.
- Threw out some unnessesary includes.
Next up: Commenting and making things look nice.
Former-commit-id: c983d1e1a2
- Removed the RewardMatrixInformationStruct since it was only used as function argument.
|- Replaced its function by the respective results of the transition parser (for det/non-det)
- Moved the MappedFile class to separate file.
- Renamed the SupportedFileEndingsEnum enum to SupportedFileEndings.
- Whats left in Parser.*:
|- A bunch of free functions for manipulation of c-style strings, a function to read the model hint and the SupportedFileEndings enum with its recognition function.
Im not exactly sure what to do with this.
- I think the name Parser.* implies a parser base class, while it is not. Therefore: Renaming.
- The c-style string manipulation functions might also needed elsewhere. Therefore: Moving them into the utility folder/namespace.
- The model hint function would fit in the AutoParser, except for its signature (low level arguments vs. high level arguments of the AutoParser). Therefore: Either moving it to the Auto Parser or leave it as helper function with the others (c-style string handling and low level OS detection)
- The SupportedLineEndings and its assorted functions either belong to the string manipulation functions or to the MappedFile class. Therefor: Moving them accordingly.
- Once all things have moved the Parser.* will be empty. Therefore: Deletion.
Next up: Trying to decide what to do with the rest of the Parser.* and doing it.
Former-commit-id: aa7417e5e1
- Devided the AutoParser.h into .h and .cpp
- The AutoParser now is a stateless class
|- This resulted in changes to the interface between the parsers and the rest of the project.
|- The main() now directly acquires a shared_ptr to an AbstractModel from the call of the AutoParser and keeps ownership of it.
|- Additionally, the division into .h and .cpp lead to a move of includes from the header to the source. This caused several tests to need some model header to be included.
|- Tests are still showing green (except those needing Gurobi, which I do not have).
Next up: Parser.h/.cpp, then comments and making things look nice.)
Former-commit-id: f59b7405e5
dehnert
masawei
PBerger
sjunges