This is a small change to add a new hook, similar to __evaluate_entry and __evaluate_exit, but for when a new frame is created from an old one. I am building an open-source pluggable framework around JSonata, and this small change will open up an enormous amount of awesome possibilities.
It is beneficial for advanced cases like modifying env with additional metadata or mutating it so that you can analyse bind/lookup calls & provide diagnostics. I have actually patially achieved this already to significant effect by hooking into the env on __evaluate_entry. However, in some cases inside the JSONata core, a new frame is created, which is then immediately bound to, and consequently that happens before it has yet to recurse through evaluate and the entry callback. That means that callback is unable to capture everything for these advance pro-mode use cases. This new callback would enable that possibility.
I accept this will be undocumented and could change in the future.
The callback is set via a Symbol, so it can not be manipulated or read inside a query. See #700, which would need to be merged first, both as approval of the general Symbol approach and because the minor TS def changes are needed from there.
This is a small change to add a new hook, similar to
__evaluate_entryand__evaluate_exit, but for when a new frame is created from an old one. I am building an open-source pluggable framework around JSonata, and this small change will open up an enormous amount of awesome possibilities.It is beneficial for advanced cases like modifying env with additional metadata or mutating it so that you can analyse bind/lookup calls & provide diagnostics. I have actually patially achieved this already to significant effect by hooking into the env on
__evaluate_entry. However, in some cases inside the JSONata core, a new frame is created, which is then immediately bound to, and consequently that happens before it has yet to recurse throughevaluateand the entry callback. That means that callback is unable to capture everything for these advance pro-mode use cases. This new callback would enable that possibility.I accept this will be undocumented and could change in the future.
The callback is set via a
Symbol, so it can not be manipulated or read inside a query. See #700, which would need to be merged first, both as approval of the generalSymbolapproach and because the minor TS def changes are needed from there.Signed-off-by: Adam Thomas adam@adam-thomas.co.uk