This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
There is the FOTS test format-integer-050: string-join(for $i in 1 to 5 return format-integer($i,''α''), ''|'') that expects: α|β|γ|δ|ε However, nothing in section 4.6.1 says why it should expect that result. Does it expect that result because: 1. It's an unwritten variant of the 'a' format token that expects the sequence a, b, c, ..., z, aa, ab, ac, ..., in which case if the value of $i were to be 25, then it would yield αα? 2. It's "any other format token, which indicates a numbering sequence in which that token represents the number 1 (one)" and furthermore "may be ·implementation-defined· lower and upper bounds on the range of numbers that can be formatted using this format token" where the lower/upper bounds in this case would be 1/24? Both cases, however, are implementation defined in which case there shouldn't be a test such as format-integer-050 having a particular expected result.
It falls under the "any other format token" rule. This test is marked with the dependency: <dependency type="format-integer-sequence" value="α" satisfied="true"/> This means that if an implementation supports a numbering sequence starting with α, it should run the test; if it does not support such a numbering sequence, it should not run the test. If there is a numbering sequence starting with α that does not proceed with β, γ, δ, ε then that alternative numbering sequence would form a reasonable alternative result for the test. The objective of the tests in this section is to try and test as many cases as possible even if the result is implementation-defined, while documenting the dependencies on implementation-defined features so that implementations can decide whether or not the test is an appropriate one to run. If an implementation chooses not to run the test because it does not support a numbering sequence starting with α, it should report the test as "n/a" in any submitted test results.
I've reclassified this as a test suite bug since it doesn't appear to raise any problem with the specification.
I've added comments to a few of the relevant tests pointing out that implementations that don't support the relevant numbering sequence should not run the test. I believe this is sufficient to close the bug; if you don't agree, please feel free to re-open.