On sequences

Achim Berndzen <achim.berndzen@xml-project.com> writes:
> By the way: Unless I miss something important, the new rules should
> also state that a output port present in one branch but missing in
> another must have @sequence=„true“. Right?

Maybe. But it would just be an extra check. If the p:when that contains
the port never gets evaluated, no one will ever check the cardinality
of its outputs.

Do we really get enough value out of @sequence=true/false to justify it?

What benefit does it provide to end users? What negative consequences
would arise if we removed it and simply said all inputs and outputs
may produce a sequence?

Steps that care, like p:xslt, where a sequence of stylesheet inputs
makes no sense, could check and raise errors.

Of course, step authors might forget to do that and only use the first
input. But maybe that’s what users would expect.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh
Lead Engineer
MarkLogic Corporation
Phone: +1 512 761 6676
www.marklogic.com