Igor Dejanović

Great! @danixeee started textX-LS recently and is actively working on it so I guess the two of you could see if and how you could integrate your efforts. I'll see...

User have full control over user classes so the obvious and preferred way to provide defaults is in the class constructor. For dynamically created textX classes we don't have other...

Yeah, it would be a change to the way it worked before. I guess that documenting the preferred way of providing the defaults might be a better approach. Something like:...

Sounds like a good idea. I think it would be relatively straightforward to implement it as an optional parameter to `model_from*` where the default value may be `None` so it...

OTOH, textX will only [set value to `None` if it is missing in the input](http://textx.github.io/textX/stable/metamodel/#auto-initialization-of-the-attributes) so the only benefit of providing sentinel value would be if user-classes are also instantiated...

Yep. Definitely should be the default behaviour. It would reduce grammar boilerplate in some cases.

Good catch. Interesting corner case. Thanks for the report.

Hi, A PEG way is to use a negative lookahead (`Not` predicate) to check that there is no keyword ahead before trying to match identifier. ```python def identifier (): return...

Hi @mabraham. Thanks for the report. It was a surprising to me as I don't recall that I did any rebuild. I increase the version number each time there is...

The tests have become part of the source tarball PyPI release since v1.9.1, so that might be the underlying reason. Please see [here](https://github.com/textX/Arpeggio/issues/51)