Opened 3 years ago
Last modified 3 years ago
#6443 new defect
Issues with profiling in OMEdit
| Reported by: | casella | Owned by: | adeas31 |
|---|---|---|---|
| Priority: | high | Milestone: | |
| Component: | OMEdit | Version: | 1.18.0-dev |
| Keywords: | Cc: | Andrea.Bartolini |
Description
Despite the improvements applied to #5804, we are still facing problems with profiling in OMEdit.
The worst problem is that, in some cases, if you turn on profiling after running some simulations on the same model, the profiling results are not loaded. Unfortunately, we were unable to reproduce this issue with a given sequence of steps. However, we believe this is due to some invalid state that the files in the working directory end up in.
For sure, we can reproduce another issue: if you run a simulation with profiling on, and then turn it off, the old profiling results keep being loaded and displayed in the debugger, also across sessions, so they may actually be months or years old. This can be quite confusing.
If we delete the simulation results directory in the working directory and start from clean conditions, then everything works fine.
We would then suggest that all *prof_.* files are deleted just before running a new simulation (both when recompiling and resimulating), to avoid weird side effects from previous simulations.
BTW, we did not understand why two prof_ files are generated even in case profiling is switched off. What information do they contain?
Ticket retargeted after milestone closed