Exercise 3: What count Evaluates To in the "staging" Workspace — Possible Solution
====================================================================

`count` evaluates to 1 while the "staging" workspace is selected.

The expression `terraform.workspace == "prod" ? 3 : 1` is a simple
two-way ternary: it checks ONE specific condition (is the current
workspace exactly equal to the string "prod"?) and has exactly two
possible outcomes -- 3 if that condition is true, 1 for literally
EVERY other case, with no distinction made between them. "staging" is
not equal to "prod," so the condition evaluates to false, and the
expression falls through to the else-branch value of 1 -- functionally
identical, from this expression's point of view, to how "dev" would
also evaluate to 1.

This is a genuine, worth-noticing gotcha: the ternary was written with
only two environments in mind (implicitly, "prod" and "everything
else"), but the moment a third workspace like "staging" is introduced,
it silently gets grouped into the same "everything else" bucket as
"dev" -- getting `count = 1`, the same as dev, even if the team's
actual intent was for staging to have its own distinct value (say, 2)
somewhere between dev's 1 and prod's 3. Nothing about running `plan`
or `apply` in the staging workspace would surface this as an error;
the configuration is perfectly valid HCL and produces a perfectly
well-formed plan -- it just silently doesn't do what a naive reading of
"we have three environments now" might assume.

The fix, if staging genuinely needs its own value, would be to replace
the two-way ternary with something that maps each workspace name to
its own value explicitly -- e.g. a lookup map keyed by
`terraform.workspace`, with a value defined for every environment
rather than an implicit "everything that isn't prod" bucket.

WHY THIS WORKS AS AN ANSWER
------------------------------
This gives the correct evaluated value (1) with the precise reasoning
(a two-way ternary has no notion of a third case), and goes further to
name the underlying real-world risk -- a new environment silently
inheriting an unrelated environment's behavior -- rather than treating
the question as pure syntax trivia.