I guess we have different working definitions of what "deterministic" means.

Wikipedia says:

A deterministic algorithm is an algorithm which, in informal terms, behaves predictably. Given a particular input, it will always produce the same output, and the underlying machine will always pass through the same sequence of states.

Perl's parser, due to it's pluggable nature, does not adhere to that definition. The output can vary for a given input, and it varies because the underlying states visited vary.

I've looked at more format definitions elsewhere and reached the same conclusion.

In this case, one could even argue that it's [...] a matter of determining whether Perl will get around to assigning meaning to some code.

When one discuses whether Perl can be parsed or not, one is not referring to the time it takes or even if it's finite. One is discussing whether the output is stable enough to study or store for future use.

Let's take C++ for example. One would consider C++ to be parseable. One can perform static analysis on C++ code. C++ code can be compiled. Yet, it's possible to introduce an infinite loop in the parser because C++'s template system is turing complete.

The difference between C++ and Perl is that C++'s parser is confined to making decision based on its input, whereas Perl's parser can make decision based on external data.