Update timings in Preesm

[antoine-morvan]

Timings in Preesm are no good:

1. The scenario editor does not allow giving different timings to different instances of an actor. This is an issue since different instances with different parameter values might have different execution time (see Scenario Editor: Issue in Timing tab when a hierarchical subgraph is instantiated multiple times. #77);
2. The scenario editor does not allow giving parameterized timings. Although instrumentation might never generated parameterized timings, the user might now the complexity of his own code (see Expression-based timings in the scenario editor #73);
3. The scenario editor cannot import instrumentation results as is: the exported CSV must be processed manually in order to import timings properly. This must be fixed (see [UI] TimingParser fails at loading timings generated via the InstrumentedC code generator #52);
4. The new instrumentation library (dump.c/h) has to be tested on C6678 (see (Codegen) Instrumented codegen has incorrect columns in analysis.csv. #95);
5. Timings seem to not be properly propagated to all steps (see Actor firing latencies set in scenario is not taken into account in the Gantt chart #121);

[antoine-morvan]

2 first bullets are fixed since 42b11ef.

[leos313]

Hi Guys,

What is new on this bug? Are the 3 last points still open?

Thanks

[antoine-morvan]

Points 1 and 2 are fixed in latest release. Alexandre H. is working on points 3 and 4. It's up to the scheduler to use the timing information (or not). The Gantt exporter should be using them as of latest release.

[ahonorat]

All related issues have been closed, and new timings seem to work (specification, instrumentation, and parsing) since bd8fe33 , thus the issue can be closed.