Provide an XML version of the stack trace

[rhdunn]

Provide an XML version of the stack trace

[ChristianGruen]

I would tend to close this issue (and keep #689 open). I suppose we’d need to introduce much stricter optimization rules in the specification to achieve reasonable results across multiple processors.

[ChristianGruen]

I’m still hesitant if we should include this function to the standard. I clearly see reasons for having a stack trace function, but we may (at this stage) need to consider numerous implementation-specific questions:

Code may be rewritten a lot by a processor before it’s eventually executed, and it’s often difficult to preserve the original location of a former expression (in particular, the column number), even if we ignore function inlining. An arbitrary example:

(: original query :)
some $s in
  (1, 2, 3)
satisfies
  $s > 1

(: rewritten query :)
((1 to 3) ! (. >= 2)) = true()

Rewritings could generally be suppressed if a stack-trace call is found in a sub expression, but this would change the way how a query is executed, and may lead to heisenbugs (and I assume the main reason for using this function is for debugging purposes).

[ndw]

The CG agreed to close this issue without action at meeting 065