You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Clearly, a big barrier to wider adoption, is the fact that many errors in the yaml file, lead to seemingly random errors when solving the model. Many of these errors could be catched and exposed by a linter, like we do in Python.
I thought that @ezgioz could extend the proof of concept so as to cover most of the errors/warnings message that we cover in Python (https://github.com/EconForge/dolo/blob/master/dolo/linter.py). This could start by listing somewhere all the errors/warnings that we must check.
There are other issues for later (I suspect @sglyon might be curious about them):
how to integrate the linter to an editor workflow (calling cmd-line might be too slow)
the proof of concept operates on the EventTree produced by YAML.jl without constructing any object. This one keeps a marker with line/column of every node. It is not the same as what is stored in model.data although we might discuss at later stages whether merging both would be a good idea (concretely replacing model.data by a structure with lines/columns informations or with the EventTree itself). There might be problems as we sometimes want to modify YAML.jl at some point.
The text was updated successfully, but these errors were encountered:
Clearly, a big barrier to wider adoption, is the fact that many errors in the yaml file, lead to seemingly random errors when solving the model. Many of these errors could be catched and exposed by a linter, like we do in Python.
I created a new branch https://github.com/EconForge/Dolo.jl/tree/linter , with a proof of concept.
I thought that @ezgioz could extend the proof of concept so as to cover most of the errors/warnings message that we cover in Python (https://github.com/EconForge/dolo/blob/master/dolo/linter.py). This could start by listing somewhere all the errors/warnings that we must check.
There are other issues for later (I suspect @sglyon might be curious about them):
model.data
although we might discuss at later stages whether merging both would be a good idea (concretely replacing model.data by a structure with lines/columns informations or with the EventTree itself). There might be problems as we sometimes want to modify YAML.jl at some point.The text was updated successfully, but these errors were encountered: