This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The proposed change to format-integer in Bug 14858 gives no hint that the expected result for this test is correct. I expected this test to mean that both primary format token and format modifier are the empty string, and thus the result would be equivalent to format-integer(1, ''). <test-case name="format-integer-061"> <description>format-integer - Error case with lone semicolon</description> <created by="Michael Kay, Saxonica" on="2012-02-28"/> <environment ref="empty"/> <test>format-integer(1, ';')</test> <result> <error code="XTDE0030"/> </result> </test-case>
The general principle, which is retained under the new rules, is that a syntactically invalid picture is an error, whereas an unsupported-but-syntactically-legal picture leads to fallback behaviour. An empty picture string falls into the category of "syntactically invalid".
Looking at the spec again, I agree with Tim. The primary format token here is a zero-length string; the chances are that this is a format-token not recognized by the implementation, and the expected behaviour is therefore fallback to a format token of "1", giving the output "1". Fixed accordingly.
Confirmed fixed.