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 XSLT 3.0 specification asserts, in section 4.7, that the Data Model spec leaves it to host languages to define limits. If it does so, it seems that it does so only by default; I can find no mention of limits in the Data Model spec. On the other hand, F+O does have a few things to say about limits, for example the required precision for date/time values. It doesn't contradict what we say in XSLT, except that it doesn't say that it's a host language responsibility. Section 4.7 says nothing about what processors should do when limits are exceeded. What happens, for example, if the limit on the length of a sequence is exceeded? We don't say whether processors are required to fail cleanly.
The Working Group was minded to do nothing. (But recognized that the issue might come up in relation to specific test cases...)