I think having region boundaries part of the code makes sense. Is the alternative having file and line number references somewhere else? They could fall out of sync all too easily.
But the open/collapsed state of the regions? That should be per-user IDE state, the last thing we want us that becoming diff noise in the repo.
I just disagree, but that’s okay. This is about presentation of code to the viewer which makes it a responsibility of the IDE and not a responsibility of the language. You’re right that the line references would need to be kept in sync, but I think that would be trivial.
I think having region boundaries part of the code makes sense. Is the alternative having file and line number references somewhere else? They could fall out of sync all too easily.
But the open/collapsed state of the regions? That should be per-user IDE state, the last thing we want us that becoming diff noise in the repo.
I just disagree, but that’s okay. This is about presentation of code to the viewer which makes it a responsibility of the IDE and not a responsibility of the language. You’re right that the line references would need to be kept in sync, but I think that would be trivial.