Finish error handling

[syntron]

Finish update of error handling in OMSession.py and ModelicaSystem.py

Based on PR #277, #273 and #270

Changes

• Always check for errors in sendExpression()
• align logging - log data only once
• cleanup

[adeas31]

Rebase this one and see if you can use getMessagesStringInternal(). We should also check if expression is getErrorString() or getMessagesStringInternal() then we should not call it twice.

[syntron]

@adeas31 would it be possible to merge #281 and #282 before that? Both would merge in this order without problems even if not rebased.

This would simplify the merge as this PR is based on these two. Furthermore, I will be limited in time till mid of next week ...

[syntron]

@adeas31 do you have a simple way / an example to force an error? I would like to check how to evaluate the output of getMessagesStringInternal()

[adeas31]

Run the following script with omc,

loadString("model M Real foo = \"str\"; annotation (uses(Modelica(version=\"4.0.0\"))); end M;");
getMessagesStringInternal();
checkModel(M);
getMessagesStringInternal();

equivalent OMPython code is,

from OMPython import OMCSessionZMQ
omc = OMCSessionZMQ()
omc.sendExpression("model M Real foo = \"str\"; annotation (uses(Modelica(version=\"4.0.0\"))); end M;")
omc.sendExpression("getMessagesStringInternal()")
omc.sendExpression("checkModel(M)")
omc.sendExpression("getMessagesStringInternal()")

The first call to getMessagesStringInternal() gives the notificaitons caused by loading model M, the second call gives the error caused by calling checkModel(M).
Note. I haven't tried running the python code but I guess it will work.

You can also see how getMessagesStringInternal() is used in OMEdit here https://github.com/OpenModelica/OpenModelica/blob/master/OMEdit/OMEditLIB/OMC/OMCProxy.cpp#L611

[adeas31]

Let me know once its ready for review.

[syntron]

Let me know once its ready for review.

@adeas31 all done including getMessagesStringInternal() in this PR

Please use the following order for review: PR #281, PR #282 (needs a simple conflict solution), PR #278

PR #279 is currently not updated as it would be on top of PR #278

Besides these, there are some smaller cleanups pending

[adeas31]

All the dependent PRs are merged. Lets rebase this one now.

[syntron]

@adeas31 ready to go ...

[adeas31]

LGTM.

[adeas31]

Do we really need to make them class members?

[adeas31]

self._re_log_entries and self._re_log_raw can be local variables here.

[syntron]

I defined them as class variables to do the initialisation once (if needed at all). Later, the compiled re struture is just reused ... However, if the expected numbers of calls to this part is low, it makes sense to run this each time.

[adeas31]

The actual re compilation is run each time, its just they are initialized to none only once. I don't think initializing them to none locally is expensive.

[syntron]

in OMCSession.py line 584ff (definition of sendExpression()):

if not self._re_log_entries:
    self._re_log_entries = re.compile(...)

if not self._re_log_raw:
    self._re_log_raw = re.compile(...)

Compilation is done only once (on first call; self._re* class variables are None); after that, the variables are set (not None) and, thus, the compiled version is used ...