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
Describe the bug nickel doc aborts with a stack overflow on a real-world 1MB Nickel schema, which doesn't look very deep (nickel doc does use recursive calls, but the stack space should be linear in the depth of the schema, roughly).
To Reproduce
I unfortunately can't provide the test case, which has been provided as a courtesy but isn't public. This issue is mostly a reminder to investigate it.
The text was updated successfully, but these errors were encountered:
I think we had the exact same problem in the LSP (not for documentation, but for elaborating e.g. completion information). One simple and stupid solution was to set a recursion limit there.
However, I don't think that would work very well for documentation, because that will result in unrolling the definition of Recursivemax_depth times, which is not what one's want. So we need a way to keep track of things that we've already processed (or are processing), and decide to not unfold those definitions. I wonder if we can reuse the logistics of thunks to do so.
Describe the bug
nickel doc
aborts with a stack overflow on a real-world 1MB Nickel schema, which doesn't look very deep (nickel doc
does use recursive calls, but the stack space should be linear in the depth of the schema, roughly).To Reproduce
I unfortunately can't provide the test case, which has been provided as a courtesy but isn't public. This issue is mostly a reminder to investigate it.
The text was updated successfully, but these errors were encountered: