Bump used Modelica language version to 3.6 for MapleSim moparser (CI)

[beutlich]

According to #4175 this should not be merged right now.

[maltelenz]

There was a decision in the monthly MAP-Lib meeting on the 14th of November, to switch to 3.6 fully, and consider for each PR that tries to use 3.6 features, whether we are ready to accept it for 4.1.0 or not.

Unfortunately it doesn't seem this was written down anywhere. It was discussed in the context of #4208.

So this should be possible to merge.

[tobolar]

@beutlich : I guess @maltelenz is better suited to review this PR. I was not present at the last MAP-Lib meeting.

[eshmoylova]

@hubertus65 recorded the following in the chat on Teams for MAP-LIB Monthly:

If we say in the documentation that we support 3.6 then we need to support that claim elsewhere.

[henrikt-ma]

@casella, I suppose you as the project leader is the right person to push the merge button on this.

Edit: My bad, I was confused and mixed up this PR with #4208.

[arunkumar-narasimhan]

@casella, this PR is all ready to merge, but seems to be failing at checks. Could you please look into it?

[beutlich]

but seems to be failing at checks.

Not any more.

Note, that merging this PR opens the door for PRs like #4231 which I'd not to have right now. Therefore I'd not like to have it merged now.

[henrikt-ma]

Note, that merging this PR opens the door for PRs like #4231 which I'd not to have right now. Therefore I'd not like to have it merged now.

This sounds like a misunderstanding to me. The door for #4231 is already open version-wise as of #4208. What is needed in order to move forward with #4231 shouldn't have anything to do with the MapleSim moparser, but what we need is an agreement with tool vendors that they welcome the adoption of the specific language feature selective model extension.

Hence, I believe this PR is fine. Moving it out of Draft state would seem like a good start.

[GallLeo]

MAP-LIB meeting 2024-02-02:
We already agreed to not open up to all Modelica 3.6 features.
But, we bumped the version in order to e.g. use figure annotations (as long as no tool vendor complains).
Therefore, syntax check should also be bumped to Modelica 3.6.

[GallLeo]

@Harisankar-Allimangalath Please back-port to release-branch v4.1.0

[beutlich]

MAP-LIB meeting 2024-02-02: We already agreed to not open up to all Modelica 3.6 features. But, we bumped the version in order to e.g. use figure annotations (as long as no tool vendor complains). Therefore, syntax check should also be bumped to Modelica 3.6.

Sorry, I cannot follow. Enabling MoLang 3.6 syntax also opens all MoLang 3.6 features (as the new break keyword).

[maltelenz]

That was what was decided last year, that we would open up for 3.6 generally. However, for each individual pull request that uses new features, tool vendors can object that they are not ready.

[Harisankar-Allimangalath]

@maltelenz @beutlich How should this be proceeded ?

[maltelenz]

@Harisankar-Allimangalath please create a pull request to backport to 4.1.0 maint branch.

[beutlich]

@Harisankar-Allimangalath please create a pull request to backport to 4.1.0 maint branch.

Really? I hope that we do not do this.

[maltelenz]

@Harisankar-Allimangalath please create a pull request to backport to 4.1.0 maint branch.

Really? I hope that we do not do this.

Could you explain why not?

[beutlich]

Because we do not want to allow new syntax features in the branched-off maint branch, do we?

[maltelenz]

The decision by MAP-LIB was to use Modelica 3.6 in 4.1.0, so yes, technically we do want to allow exactly that.

In practice, tool vendors will object if a pull request using break appears on the maint branch for 4.1.0.

[Harisankar-Allimangalath]

@AHaumer should I backport this? This delays the tool vendor testing.

[beutlich]

This delays the tool vendor testing.

I am afraid I cannot follow. This a CI feature and tool vendors can test with or w/o it already.

[Harisankar-Allimangalath]

@beutlich ok , so once #4299 and #4298 is merged I can tagout the beta tag right ?

[maltelenz]

@Harisankar-Allimangalath The real blocker for tool vendor testing (at least ours) is that we don't have new reference results.

[beutlich]

@beutlich ok , so once #4299 and #4298 is merged I can tagout the beta tag right ?

IMO it is not yet there. The changes since alpha.1 are hardly relevant so far whereas the actual open issues haven't be addressed yet.

[beutlich]

The real blocker for tool vendor testing (at least ours) is that we don't have new reference results.

Unblocking -> Results by Dymola 2024x of MSL v4.1.0-beta.1 (with serveral necessary changes) are here.

[maltelenz]

The real blocker for tool vendor testing (at least ours) is that we don't have new reference results.

Unblocking -> Results by Dymola 2024x of MSL v4.1.0-beta.1 (with serveral necessary changes) are here.

While the effort is appreciated, I was hoping for these things before looking at correctness failures for System Modeler:

• Only updating the reference results for models where we have good reason (model changes)
• At least a categorization of the current failures in the regression testing with Dymola, which I think @GallLeo and @casella started on?

[beutlich]

While the effort is appreciated, I was hoping for these things before looking at correctness failures for System Modeler:

* Only updating the reference results for models where we have good reason (model changes)

Uff, not sure if this can be tackled at all while also updating the simulation tool.

* At least a categorization of the current failures in the regression testing with Dymola, which I think @GallLeo and @casella started on?

Well, #4296 only touched four models and there was no vibrant discussion going on which I would have expected before releasing a beta release.

[maltelenz]

Well, #4296 only touched four models and there was no vibrant discussion going on which I would have expected before releasing a beta release.

Right. I don't know what the intention is with the beta, but if it is a signal that it is ready for tool vendor testing, it's clearly not the case yet :(

[beutlich]

Yes, intention of the beta is to initiate tool vendor testing. However with reference results being avialble only now, the process is already broken. Still, you should not be blocked anymore.